MOTIVATION: One of the core problems in the analysis of protein tandem mass spectrometry data is the peptide assignment problem: determining, for each observed spectrum, the peptide sequence that was responsible for generating the spectrum. Two primary classes of methods are used to solve this problem: database search and de novo peptide sequencing. State-of-the-art methods for de novo sequencing use machine learning methods, whereas most database search engines use hand-designed score functions to evaluate the quality of a match between an observed spectrum and a candidate peptide from the database. We hypothesized that machine learning models for de novo sequencing implicitly learn a score function that captures the relationship between peptides and spectra, and thus may be re-purposed as a score function for database search. Because this score function is trained from massive amounts of mass spectrometry data, it could potentially outperform existing, hand-designed database search tools. RESULTS: To test this hypothesis, we re-engineered Casanovo, which has been shown to provide state-of-the-art de novo sequencing capabilities, to assign scores to given peptide-spectrum pairs. We then evaluated the statistical power of this Casanovo score function, Casanovo-DB, to detect peptides on a benchmark of three mass spectrometry runs from three different species. In addition, we show that re-scoring with the Percolator post-processor benefits Casanovo-DB more than other score functions, further increasing the number of detected peptides.

MOTIVATION: High-resolution Hi-C contact matrices reveal the detailed three-dimensional architecture of the genome, but high-coverage experimental Hi-C data are expensive to generate. Simultaneously, chromatin structure analyses struggle with extremely sparse contact matrices. To address this problem, computational methods to enhance low-coverage contact matrices have been developed, but existing methods are largely based on resolution enhancement methods for natural images and hence often employ models that do not distinguish between biologically meaningful contacts, such as loops and other stochastic contacts. RESULTS: We present Capricorn, a machine learning model for Hi-C resolution enhancement that incorporates small-scale chromatin features as additional views of the input Hi-C contact matrix and leverages a diffusion probability model backbone to generate a high-coverage matrix. We show that Capricorn outperforms the state of the art in a cross-cell-line setting, improving on existing methods by 17% in mean squared error and 26% in F1 score for chromatin loop identification from the generated high-coverage data. We also demonstrate that Capricorn performs well in the cross-chromosome setting and cross-chromosome, cross-cell-line setting, improving the downstream loop F1 score by 14% relative to existing methods. We further show that our multiview idea can also be used to improve several existing methods, HiCARN and HiCNN, indicating the wide applicability of this approach. Finally, we use DNA sequence to validate discovered loops and find that the fraction of CTCF-supported loops from Capricorn is similar to those identified from the high-coverage data. Capricorn is a powerful Hi-C resolution enhancement method that enables scientists to find chromatin features that cannot be identified in the low-coverage contact matrix. AVAILABILITY AND IMPLEMENTATION: Implementation of Capricorn and source code for reproducing all figures in this paper are available at https://github.com/CHNFTQ/Capricorn.

MOTIVATION: The Basic Local Alignment Search Tool, BLAST, is an indispensable tool for genomic research. BLAST has established itself as the canonical tool for sequence similarity search in large part thanks to its meaningful statistical analysis. Specifically, BLAST reports the E-value of each reported alignment, which is defined as the expected number of optimal local alignments that will score at least as high as the observed alignment score, assuming that the query and the database sequences are randomly generated. RESULTS: Here, we critically evaluate the E-values provided by the standard protein BLAST (blastp), showing that they can be at times significantly conservative while at others too liberal. We offer an alternative approach based on generating a small sample from the null distribution of random optimal alignments, and testing whether the observed alignment score is consistent with it. In contrast with blastp, our significance analysis seems valid, in the sense that it did not deliver inflated significance estimates in any of our extensive experiments. Moreover, although our method is slightly conservative, it is often significantly less so than the blastp E-value. Indeed, in cases where blastp's analysis is valid (i.e., not too liberal), our approach seems to deliver a greater number of correct alignments. One advantage of our approach is that it works with any reasonable choice of substitution matrix and gap penalties, avoiding blastp's limited options of matrices and penalties. In addition, we can formulate the problem using a canonical family-wise error rate control setup, thereby dispensing with E-values, which can at times be difficult to interpret. AVAILABILITY AND IMPLEMENTATION: The Apache licensed source code is available at https://github.com/batmen-lab/SGPvalue.

MOTIVATION: Interpretation of newly acquired mass spectrometry data can be improved by identifying, from an online repository, previous mass spectrometry runs that resemble the new data. However, this retrieval task requires computing the similarity between an arbitrary pair of mass spectrometry runs. This is particularly challenging for runs acquired using different experimental protocols. RESULTS: We propose a method, MS1Connect, that calculates the similarity between a pair of runs by examining only the intact peptide (MS1) scans, and we show evidence that the MS1Connect score is accurate. Specifically, we show that MS1Connect outperforms several baseline methods on the task of predicting the species from which a given proteomics sample originated. In addition, we show that MS1Connect scores are highly correlated with similarities computed from fragment (MS2) scans, even though these data are not used by MS1Connect. AVAILABILITY AND IMPLEMENTATION: The MS1Connect software is available at https://github.com/bmx8177/MS1Connect. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: Modality matching in single-cell omics data analysis-i.e. matching cells across datasets collected using different types of genomic assays-has become an important problem, because unifying perspectives across different technologies holds the promise of yielding biological and clinical discoveries. However, single-cell dataset sizes can now reach hundreds of thousands to millions of cells, which remain out of reach for most multimodal computational methods. RESULTS: We propose LSMMD-MA, a large-scale Python implementation of the MMD-MA method for multimodal data integration. In LSMMD-MA, we reformulate the MMD-MA optimization problem using linear algebra and solve it with KeOps, a CUDA framework for symbolic matrix computation in Python. We show that LSMMD-MA scales to a million cells in each modality, two orders of magnitude greater than existing implementations. AVAILABILITY AND IMPLEMENTATION: LSMMD-MA is freely available at https://github.com/google-research/large_scale_mmdma and archived at https://doi.org/10.5281/zenodo.8076311.

MOTIVATION: We address the challenge of inferring a consensus 3D model of genome architecture from Hi-C data. Existing approaches most often rely on a two-step algorithm: first, convert the contact counts into distances, then optimize an objective function akin to multidimensional scaling (MDS) to infer a 3D model. Other approaches use a maximum likelihood approach, modeling the contact counts between two loci as a Poisson random variable whose intensity is a decreasing function of the distance between them. However, a Poisson model of contact counts implies that the variance of the data is equal to the mean, a relationship that is often too restrictive to properly model count data. RESULTS: We first confirm the presence of overdispersion in several real Hi-C datasets, and we show that the overdispersion arises even in simulated datasets. We then propose a new model, called Pastis-NB, where we replace the Poisson model of contact counts by a negative binomial one, which is parametrized by a mean and a separate dispersion parameter. The dispersion parameter allows the variance to be adjusted independently from the mean, thus better modeling overdispersed data. We compare the results of Pastis-NB to those of several previously published algorithms, both MDS-based and statistical methods. We show that the negative binomial inference yields more accurate structures on simulated data, and more robust structures than other models across real Hi-C replicates and across different resolutions. AVAILABILITY AND IMPLEMENTATION: A Python implementation of Pastis-NB is available at https://github.com/hiclib/pastis under the BSD license. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Understanding the impact of regulatory variants on complex phenotypes is a significant challenge because the genes and pathways that are targeted by such variants and the cell type context in which regulatory variants operate are typically unknown. Cell-type-specific long-range regulatory interactions that occur between a distal regulatory sequence and a gene offer a powerful framework for examining the impact of regulatory variants on complex phenotypes. However, high-resolution maps of such long-range interactions are available only for a handful of cell types. Furthermore, identifying specific gene subnetworks or pathways that are targeted by a set of variants is a significant challenge. We have developed L-HiC-Reg, a Random Forests regression method to predict high-resolution contact counts in new cell types, and a network-based framework to identify candidate cell-type-specific gene networks targeted by a set of variants from a genome-wide association study (GWAS). We applied our approach to predict interactions in 55 Roadmap Epigenomics Mapping Consortium cell types, which we used to interpret regulatory single nucleotide polymorphisms (SNPs) in the NHGRI-EBI GWAS catalogue. Using our approach, we performed an in-depth characterization of fifteen different phenotypes including schizophrenia, coronary artery disease (CAD) and Crohn's disease. We found differentially wired subnetworks consisting of known as well as novel gene targets of regulatory SNPs. Taken together, our compendium of interactions and the associated network-based analysis pipeline leverages long-range regulatory interactions to examine the context-specific impact of regulatory variation in complex phenotypes.

Single cell ATAC-seq (scATAC-seq) enables the mapping of regulatory elements in fine-grained cell types. Despite this advance, analysis of the resulting data is challenging, and large scale scATAC-seq data are difficult to obtain and expensive to generate. This motivates a method to leverage information from previously generated large scale scATAC-seq or scRNA-seq data to guide our analysis of new scATAC-seq datasets. We analyze scATAC-seq data using latent Dirichlet allocation (LDA), a Bayesian algorithm that was developed to model text corpora, summarizing documents as mixtures of topics defined based on the words that distinguish the documents. When applied to scATAC-seq, LDA treats cells as documents and their accessible sites as words, identifying "topics" based on the cell type-specific accessible sites in those cells. Previous work used uniform symmetric priors in LDA, but we hypothesized that nonuniform matrix priors generated from LDA models trained on existing data sets may enable improved detection of cell types in new data sets, especially if they have relatively few cells. In this work, we test this hypothesis in scATAC-seq data from whole C. elegans nematodes and SHARE-seq data from mouse skin cells. We show that nonsymmetric matrix priors for LDA improve our ability to capture cell type information from small scATAC-seq datasets.

If you are working in the field of computational biology, then hopefully you are familiar with the excitement associated with coming up with a new idea and thinking about how to follow up on it.Maybe the idea came from a talk you heard at a conference, a paper you read, or a conversation with a colleague.Regardless, your brain is now abuzz with how this idea will be implemented and what data you'll need to validate it.Ultimately, if your idea pans out, perhaps it will lead to profound scientific insights, a high-impact paper, and a widely used software tool.But for now, it's just an idea in your head.How do you begin to bring your new idea to fruition?This is, of course, the core of the scientific method-transforming an idea (or hypothesis) into discoveries.Hence, your success as a scientist depends strongly on your ability to efficiently and effectively carry out such transformations.Here, I focus on the very first few steps of that transformation, providing some general rules that can help start your new project in the right direction.Some of these rules may be considered variations on the questions asked in the Heilmeier catechism (https://www.darpa. mil/work-with-us/heilmeier-catechism), which is a set of questions that the US Defense Advanced Projects Agency uses to evaluate potential research projects. Rule 1: Write it downMany computationally inclined people prefer writing computer code to writing English.Unfortunately, if you get excited by an idea and immediately dive into implementing and validating it, then you can end up pursuing a half-baked idea that, on second thought, really doesn't make sense.The act of writing your idea down will inevitably clarify it for you.I typically think of this writeup as a first draft of the introduction of the paper you eventually aim to write.Rules 2 to 8 describe this writeup in more detail. Rule 2: Define the problem you want to solveIt is important up front to clearly delineate the scope and precise nature of the problem your work will address.If you are proposing a novel computational method, you will want to specify precisely what types of inputs are expected and what the resulting output will be.

Tandem mass spectrometry is the only high-throughput method for analyzing the protein content of complex biological samples and is thus the primary technology driving the growth of the field of proteomics. A key outstanding challenge in this field involves identifying the sequence of amino acids -the peptide- responsible for generating each observed spectrum, without making use of prior knowledge in the form of a peptide sequence database. Although various machine learning methods have been developed to address this de novo sequencing problem, challenges that arise when modeling tandem mass spectra have led to complex models that combine multiple neural networks and post-processing steps. We propose a simple yet powerful method for de novo peptide sequencing, Casanovo, that uses a transformer framework to map directly from a sequence of observed peaks (a mass spectrum) to a sequence of amino acids (a peptide). Our experiments show that Casanovo achieves state-of-the-art performance on a benchmark dataset using a standard cross-species evaluation framework which involves testing with spectra with never-before-seen peptide labels. Casanovo not only achieves superior performance but does so at a fraction of the model complexity and inference time required by other methods.

The problem of decomposing a graph flow into a small set of paths has a wide range of applications, including transcriptome assembly and routing in data networks. A standard formulation is the sparsest flow decomposition problem, which is known to be NP-hard. In this work, we consider a multi-sample variant of this problem, motivated by the problem of identifying and quantifying proteoforms from mass spectrometry data, where multiple views of the graph can be obtained from multiple biological samples. We derive necessary conditions for the set of samples to unambiguously determine the ground truth set of paths, and we design an algorithm with matching sufficient conditions for a large class of problem instances, making our algorithm information optimal for this class of problem instances. The necessary conditions, combined with a probabilistic model for sample generation, yield a characterization of the number of samples needed for unambiguous recovery of the underlying paths. We analyze the algorithm’s performance on flow data simulated on peptide graphs from real mass spectrometry data.

MOTIVATION: Target-decoy competition (TDC) is a commonly used method for false discovery rate (FDR) control in the analysis of tandem mass spectrometry data. This type of competition-based FDR control has recently gained significant popularity in other fields after Barber and Candès laid its theoretical foundation in a more general setting that included the feature selection problem. In both cases, the competition is based on a head-to-head comparison between an (observed) target score and a corresponding decoy (knockoff) score. However, the effectiveness of TDC depends on whether the data are homogeneous, which is often not the case: in many settings, the data consist of groups with different score profiles or different proportions of true nulls. In such cases, applying TDC while ignoring the group structure often yields imbalanced lists of discoveries, where some groups might include relatively many false discoveries and other groups include relatively very few. On the other hand, as we show, the alternative approach of applying TDC separately to each group does not rigorously control the FDR. RESULTS: We developed Group-walk, a procedure that controls the FDR in the target-decoy/knockoff setting while taking into account a given group structure. Group-walk is derived from the recently developed AdaPT-a general framework for controlling the FDR with side-information. We show using simulated and real datasets that when the data naturally divide into groups with different characteristics Group-walk can deliver consistent power gains that in some cases are substantial. These groupings include the precursor charge state (4% more discovered peptides at 1% FDR threshold), the peptide length (3.6% increase) and the mass difference due to modifications (26% increase). AVAILABILITY AND IMPLEMENTATION: Group-walk is available at https://cran.r-project.org/web/packages/groupwalk/index.html. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: A wide variety of experimental methods are available to characterize different properties of single cells in a complex biosample. However, because these measurement techniques are typically destructive, researchers are often presented with complementary measurements from disjoint subsets of cells, providing a fragmented view of the cell's biological processes. This creates a need for computational tools capable of integrating disjoint multi-omics data. Because different measurements typically do not share any features, the problem requires the integration to be done in unsupervised fashion. Recently, several methods have been proposed that project the cell measurements into a common latent space and attempt to align the corresponding low-dimensional manifolds. RESULTS: In this study, we present an approach, Synmatch, which produces a direct matching of the cells between modalities by exploiting information about neighborhood structure in each modality. Synmatch relies on the intuition that cells which are close in one measurement space should be close in the other as well. This allows us to formulate the matching problem as a constrained supermodular optimization problem over neighborhood structures that can be solved efficiently. We show that our approach successfully matches cells in small real multi-omics datasets and performs favorably when compared with recently published state-of-the-art methods. Further, we demonstrate that Synmatch is capable of scaling to large datasets of thousands of cells. AVAILABILITY AND IMPLEMENTATION: The Synmatch code and data used in this manuscript are available at https://github.com/Noble-Lab/synmatch.

Saliency methods can make deep neural network predictions more interpretable by identifying a set of critical features in an input sample, such as pixels that contribute most strongly to a prediction made by an image classifier. Unfortunately, recent evidence suggests that many saliency methods poorly perform, especially in situations where gradients are saturated, inputs contain adversarial perturbations, or predictions rely upon inter-feature dependence. To address these issues, we propose a framework, DANCE, which improves the robustness of saliency methods by following a two-step procedure. First, we introduce a perturbation mechanism that subtly varies the input sample without changing its intermediate representations. Using this approach, we can gather a corpus of perturbed ("decoy") data samples while ensuring that the perturbed and original input samples follow similar distributions. Second, we compute saliency maps for the decoy samples and propose a new method to aggregate saliency maps. With this design, we offset influence of gradient saturation. From a theoretical perspective, we show that the aggregated saliency map not only captures inter-feature dependence but, more importantly, is robust against previously described adversarial perturbation methods. Our empirical results suggest that, both qualitatively and quantitatively, DANCE outperforms existing methods in a variety of application domains.

A common workflow in single-cell RNA-seq analysis is to project the data to a latent space, cluster the cells in that space, and identify sets of marker genes that explain the differences among the discovered clusters. A primary drawback to this three-step procedure is that each step is carried out independently, thereby neglecting the effects of the nonlinear embedding and inter-gene dependencies on the selection of marker genes. Here we propose an integrated deep learning framework, Adversarial Clustering Explanation (ACE), that bundles all three steps into a single workflow. The method thus moves away from the notion of "marker genes" to instead identify a panel of explanatory genes. This panel may include genes that are not only enriched but also depleted relative to other cell types, as well as genes that exhibit differences between closely related cell types. Empirically, we demonstrate that ACE is able to identify gene panels that are both highly discriminative and nonredundant, and we demonstrate the applicability of ACE to an image recognition task.

MOTIVATION: Hi-C is the most widely used assay for investigating genome-wide 3D organization of chromatin. When working with Hi-C data, it is often useful to calculate the similarity between contact matrices in order to assess experimental reproducibility or to quantify relationships among Hi-C data from related samples. The HiCRep algorithm has been widely adopted for this task, but the existing R implementation suffers from run time limitations on high-resolution Hi-C data or on large single-cell Hi-C datasets. RESULTS: We introduce a Python implementation of HiCRep and demonstrate that it is much faster and consumes much less memory than the existing R implementation. Furthermore, we give examples of HiCRep's ability to accurately distinguish replicates from non-replicates and to reveal cell type structure among collections of Hi-C data. AVAILABILITY AND IMPLEMENTATION: HiCRep.py and its documentation are available with a GPL license at https://github.com/Noble-Lab/hicrep. The software may be installed automatically using the pip package installer. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: Tandem mass spectrometry data acquired using data independent acquisition (DIA) is challenging to interpret because the data exhibits complex structure along both the mass-to-charge (m/z) and time axes. The most common approach to analyzing this type of data makes use of a library of previously observed DIA data patterns (a 'spectral library'), but this approach is expensive because the libraries do not typically generalize well across laboratories. RESULTS: Here, we propose DIAmeter, a search engine that detects peptides in DIA data using only a peptide sequence database. Although some existing library-free DIA analysis methods (i) support data generated using both wide and narrow isolation windows, (ii) detect peptides containing post-translational modifications, (iii) analyze data from a variety of instrument platforms and (iv) are capable of detecting peptides even in the absence of detectable signal in the survey (MS1) scan, DIAmeter is the only method that offers all four capabilities in a single tool. AVAILABILITY AND IMPLEMENTATION: The open source, Apache licensed source code is available as part of the Crux mass spectrometry analysis toolkit (http://crux.ms). SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: Successful science often involves not only performing experiments well, but also choosing well among many possible experiments. In a hypothesis generation setting, choosing an experiment well means choosing an experiment whose results are interesting or novel. In this work, we formalize this selection procedure in the context of genomics and epigenomics data generation. Specifically, we consider the task faced by a scientific consortium such as the National Institutes of Health ENCODE Consortium, whose goal is to characterize all of the functional elements in the human genome. Given a list of possible cell types or tissue types ('biosamples') and a list of possible high-throughput sequencing assays, where at least one experiment has been performed in each biosample and for each assay, we ask 'Which experiments should ENCODE perform next?' RESULTS: We demonstrate how to represent this task as a submodular optimization problem, where the goal is to choose a panel of experiments that maximize the facility location function. A key aspect of our approach is that we use imputed data, rather than experimental data, to directly answer the posed question. We find that, across several evaluations, our method chooses a panel of experiments that span a diversity of biochemical activity. Finally, we propose two modifications of the facility location function, including a novel submodular-supermodular function, that allow incorporation of domain knowledge or constraints into the optimization procedure. AVAILABILITY AND IMPLEMENTATION: Our method is available as a Python package at https://github.com/jmschrei/kiwano and can be installed using the command pip install kiwano. The source code used here and the similarity matrix can be found at http://doi.org/10.5281/zenodo.3708538. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

We present apricot, an open source Python package for selecting representative subsets from large data sets using submodular optimization. The package implements several efficient greedy selection algorithms that offer strong theoretical guarantees on the quality of the selected set. Additionally, several submodular set functions are implemented, including facility location, which is broadly applicable but requires memory quadratic in the number of examples in the data set, and a feature-based function that is less broadly applicable but can scale to millions of examples. Apricot is extremely efficient, using both algorithmic speedups such as the lazy greedy algorithm and memoization as well as code optimization using numba. We demonstrate the use of subset selection by training machine learning models to comparable accuracy using either the full data set or a representative subset thereof. This paper presents an explanation of submodular selection, an overview of the features in apricot, and applications to two data sets.

Single-cell Hi-C (scHi-C) interrogates genome-wide chromatin interaction in individual cells, allowing us to gain insights into 3D genome organization. However, the extremely sparse nature of scHi-C data poses a significant barrier to analysis, limiting our ability to tease out hidden biological information. In this work, we approach this problem by applying topic modeling to scHi-C data. Topic modeling is well-suited for discovering latent topics in a collection of discrete data. For our analysis, we generate nine different single-cell combinatorial indexed Hi-C (sci-Hi-C) libraries from five human cell lines (GM12878, H1Esc, HFF, IMR90, and HAP1), consisting over 19,000 cells. We demonstrate that topic modeling is able to successfully capture cell type differences from sci-Hi-C data in the form of "chromatin topics." We further show enrichment of particular compartment structures associated with locus pairs in these topics.

The 3D organization of the genome plays a key role in many cellular processes, such as gene regulation, differentiation, and replication. Assays like Hi-C measure DNA-DNA contacts in a high-throughput fashion, and inferring accurate 3D models of chromosomes can yield insights hidden in the raw data. For example, structural inference can account for noise in the data, disambiguate the distinct structures of homologous chromosomes, orient genomic regions relative to nuclear landmarks, and serve as a framework for integrating other data types. Although many methods exist to infer the 3D structure of haploid genomes, inferring a diploid structure from Hi-C data is still an open problem. Indeed, the diploid case is very challenging, because Hi-C data typically does not distinguish between homologous chromosomes. We propose a method to infer 3D diploid genomes from Hi-C data. We demonstrate the accuracy of the method on simulated data, and we also use the method to infer 3D structures for mouse chromosome X, confirming that the active homolog exhibits a bipartite structure, whereas the active homolog does not.

International audience

MOTIVATION: Post-translational modifications (PTMs) of proteins are associated with many significant biological functions and can be identified in high throughput using tandem mass spectrometry. Many PTMs are associated with short sequence patterns called 'motifs' that help localize the modifying enzyme. Accordingly, many algorithms have been designed to identify these motifs from mass spectrometry data. Accurate statistical confidence estimates for discovered motifs are critically important for proper interpretation and in the design of downstream experimental validation. RESULTS: We describe a method for assigning statistical confidence estimates to PTM motifs, and we demonstrate that this method provides accurate P-values on both simulated and real data. Our methods are implemented in MoMo, a software tool for discovering motifs among sets of PTMs that we make available as a web server and as downloadable source code. MoMo re-implements the two most widely used PTM motif discovery algorithms-motif-x and MoDL-while offering many enhancements. Relative to motif-x, MoMo offers improved statistical confidence estimates and more accurate calculation of motif scores. The MoMo web server offers more proteome databases, more input formats, larger inputs and longer running times than the motif-x web server. Finally, our study demonstrates that the confidence estimates produced by motif-x are inaccurate. This inaccuracy stems in part from the common practice of drawing 'background' peptides from an unshuffled proteome database. Our results thus suggest that many of the papers that use motif-x to find motifs may be reporting results that lack statistical support. AVAILABILITY AND IMPLEMENTATION: The MoMo web server and source code are provided at http://meme-suite.org. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Vishnepolsky and Pirtskhalava offer two primary critiques of our previous work. We begin by responding to the second of the two critiques, because it involves a clear error on our part. Indeed, as Vishnepolsky and Pirtskhalava point out, the script that we used to calculate results for their tool, DBAASP, included a bug such that the number of true positives should be 306 instead of 121 in Table 1 of our paper. The corrected version of Table 1 is included in this response. We apologize for this error and thank Vishnepolsky and Pirtskhalava for pointing it out. Summary of threshold-based results for the DAMPD benchmark dataset Summary of threshold-based results for the DAMPD benchmark dataset The second critique concerns the inclusion in our benchmark of peptides that are explicitly disallowed as input by some of the tools. Specifically, several of the tools do not make predictions for peptides with lengths >100 amino acid and also do not handle peptides that contain non-canonical amino acids (e.g. ‘X’ as an ambiguity code). On this point, we disagree with Vishnepolsky and Pirtskhalava. Restricting the benchmark dataset to the task of predicting only with respect to short peptides makes the benchmark unrealistically easy. We did acknowledge in the original manuscript that several of the tested tools were not designed to handle longer sequences, and we even explicitly included results from a version of the benchmark in which longer peptides are included (Fig. 4 in the original paper). But we believe that a proper comparison should include realistic peptides, and as we discussed in the manuscript, some actual antimicrobial peptides (AMPs) are longer than 100 amino acids. Similarly, though we did not point this out explicitly, some AMPs in UniProt contain non-canonical amino acids. A tool that can handle these cases is, in our opinion, more generally useful than a tool that cannot handle them. We do agree with Vishnepolsky and Pirtskhalava that explicitly pointing out, in the original manuscript, the impact of the non-canonical amino acids on our results would have been a good idea. Vishnepolsky and Pirtskhalava also suggest that evaluation of their DBAASP tool should be done by taking into account C-terminal amidation. However, the DAMPD database from which we generated our benchmark does not include indications of C-terminal amidation. We therefore made the assumption that the peptides were not modified in this way. Given that the C-terminal amidation information is not provided, we are not sure how Vishnepolsky and Pirtskhalava were able to perform this analysis and achieve the stated improvement in sensitivity of 5% because the details of their analysis were omitted. Finally, Vishnepolsky and Pirtskhalava carry out an additional comparison using a list of AMPs from the DAMPD dataset (L-DAMPD). The results in their Table 4 clearly suggest that CAMPR3(RF) outperforms DBAASP in terms of sensitivity and balanced accuracy. A further analysis using the Delong test, which compares the difference in receiver operating characteristic area for paired tools, would be required to show which method is better in a rank-based setting. These results from the DAMPD dataset, coupled with the fact that CAMPR3(RF) not only predicts whether a peptide is AMP or non-AMP but also provides an associated probability, support our initial claim that CAMPR3(RF) is a generally useful AMP prediction tool. This result is not inconsistent with the possibility, as claimed by Vishnepolsky and Pirtskhalava, that DBAASP works better for some subclasses of AMPs than others. This work was supported by King Abdullah International Medical Research Center [grant #RC16/089]; and by National Institutes of Health award [P41 GM103533]. Conflict of Interest: none declared.

Empirical evidence suggests that the malaria parasite Plasmodium falciparum employs a broad range of mechanisms to regulate gene transcription throughout the organism's complex life cycle. To better understand this regulatory machinery, we assembled a rich collection of genomic and epigenomic data sets, including information about transcription factor (TF) binding motifs, patterns of covalent histone modifications, nucleosome occupancy, GC content, and global 3D genome architecture. We used these data to train machine learning models to discriminate between high-expression and low-expression genes, focusing on three distinct stages of the red blood cell phase of the Plasmodium life cycle. Our results highlight the importance of histone modifications and 3D chromatin architecture in Plasmodium transcriptional regulation and suggest that AP2 transcription factors may play a limited regulatory role, perhaps operating in conjunction with epigenetic factors.

MOTIVATION: Identification of spectra produced by a shotgun proteomics mass spectrometry experiment is commonly performed by searching the observed spectra against a peptide database. The heart of this search procedure is a score function that evaluates the quality of a hypothesized match between an observed spectrum and a theoretical spectrum corresponding to a particular peptide sequence. Accordingly, the success of a spectrum analysis pipeline depends critically upon this peptide-spectrum score function. We develop peptide-spectrum score functions that compute the maximum value of a submodular function under $m$ m matroid constraints. We call this procedure a submodular generalized matching (SGM) since it generalizes bipartite matching. We use a greedy algorithm to compute maximization, which can achieve a solution whose objective is guaranteed to be at least $\frac{1}{1+m}$ 1 1 + m of the true optimum. The advantage of the SGM framework is that known long-range properties of experimental spectra can be modeled by designing suitable submodular functions and matroid constraints. Experiments on four data sets from various organisms and mass spectrometry platforms show that the SGM approach leads to significantly improved performance compared to several state-of-the-art methods. Supplementary information, C++ source code, and data sets can be found at https://melodi-lab.github.io/SGM.

We study the problem of maximizing deep submodular functions (DSFs) subject to a matroid constraint. DSFs are an expressive class of submodular functions that include, as strict subfamilies, the facility location, weighted coverage, and sums of concave composed with modular functions. We use a strategy similar to the continuous greedy approach, but we show that the multilinear extension of any DSF has a natural and computationally attainable concave relaxation that we can optimize using gradient ascent. Our results show a guarantee of $\max_{0<\delta<1}(1-\epsilon-\delta-e^{-\delta^2\Omega(k)})$ with a running time of $O(\nicefrac{n^2}{\epsilon^2})$ plus time for pipage rounding to recover a discrete solution, where $k$ is the rank of the matroid constraint. This bound is often better than the standard $1-1/e$ guarantee of the continuous greedy algorithm, but runs much faster. Our bound also holds even for fully curved ($c=1$) functions where the guarantee of $1-c/e$ degenerates to $1-1/e$ where $c$ is the curvature of $f$. We perform computational experiments that support our theoretical results.

Deep learning has become increasingly popular in both supervised and unsupervised machine learning thanks to its outstanding empirical performance. However, because of their intrinsic complexity, most deep learning methods are largely treated as black box tools with little interpretability. Even though recent attempts have been made to facilitate the interpretability of deep neural networks (DNNs), existing methods are susceptible to noise and lack of robustness. Therefore, scientists are justifiably cautious about the reproducibility of the discoveries, which is often related to the interpretability of the underlying statistical models. In this paper, we describe a method to increase the interpretability and reproducibility of DNNs by incorporating the idea of feature selection with controlled error rate. By designing a new DNN architecture and integrating it with the recently proposed knockoffs framework, we perform feature selection with a controlled error rate, while maintaining high power. This new method, DeepPINK (Deep feature selection using Paired-Input Nonlinear Knockoffs), is applied to both simulated and real data sets to demonstrate its empirical utility.

Summary: Segway performs semi-automated genome annotation, discovering joint patterns across multiple genomic signal datasets. We discuss a major new version of Segway and highlight its ability to model data with substantially greater accuracy. Major enhancements in Segway 2.0 include the ability to model data with a mixture of Gaussians, enabling capture of arbitrarily complex signal distributions, and minibatch training, leading to better learned parameters. Availability and implementation: Segway and its source code are freely available for download at http://segway.hoffmanlab.org. We have made available scripts (https://doi.org/10.5281/zenodo.802939) and datasets (https://doi.org/10.5281/zenodo.802906) for this paper's analysis. Contact: [email protected]. Supplementary information: Supplementary data are available at Bioinformatics online.

Motivation: Single-cell Hi-C (scHi-C) data promises to enable scientists to interrogate the 3D architecture of DNA in the nucleus of the cell, studying how this structure varies stochastically or along developmental or cell-cycle axes. However, Hi-C data analysis requires methods that take into account the unique characteristics of this type of data. In this work, we explore whether methods that have been developed previously for the analysis of bulk Hi-C data can be applied to scHi-C data. We apply methods designed for analysis of bulk Hi-C data to scHi-C data in conjunction with unsupervised embedding. Results: We find that one of these methods, HiCRep, when used in conjunction with multidimensional scaling (MDS), strongly outperforms three other methods, including a technique that has been used previously for scHi-C analysis. We also provide evidence that the HiCRep/MDS method is robust to extremely low per-cell sequencing depth, that this robustness is improved even further when high-coverage and low-coverage cells are projected together, and that the method can be used to jointly embed cells from multiple published datasets.

Motivation: The three-dimensional organization of chromatin plays a critical role in gene regulation and disease. High-throughput chromosome conformation capture experiments such as Hi-C are used to obtain genome-wide maps of three-dimensional chromatin contacts. However, robust estimation of data quality and systematic comparison of these contact maps is challenging due to the multi-scale, hierarchical structure of chromatin contacts and the resulting properties of experimental noise in the data. Measuring concordance of contact maps is important for assessing reproducibility of replicate experiments and for modeling variation between different cellular contexts. Results: We introduce a concordance measure called DIfferences between Smoothed COntact maps (GenomeDISCO) for assessing the similarity of a pair of contact maps obtained from chromosome conformation capture experiments. The key idea is to smooth contact maps using random walks on the contact map graph, before estimating concordance. We use simulated datasets to benchmark GenomeDISCO's sensitivity to different types of noise that affect chromatin contact maps. When applied to a large collection of Hi-C datasets, GenomeDISCO accurately distinguishes biological replicates from samples obtained from different cell types. GenomeDISCO also generalizes to other chromosome conformation capture assays, such as HiChIP. Availability and implementation: Software implementing GenomeDISCO is available at https://github.com/kundajelab/genomedisco. Supplementary information: Supplementary data are available at Bioinformatics online.

MOTIVATION: Antimicrobial peptides (AMPs) are innate immune molecules that exhibit activities against a range of microbes, including bacteria, fungi, viruses and protozoa. Recent increases in microbial resistance against current drugs has led to a concomitant increase in the need for novel antimicrobial agents. Over the last decade, a number of AMP prediction tools have been designed and made freely available online. These AMP prediction tools show potential to discriminate AMPs from non-AMPs, but the relative quality of the predictions produced by the various tools is difficult to quantify. RESULTS: We compiled two sets of AMP and non-AMP peptides, separated into three categories-antimicrobial, antibacterial and bacteriocins. Using these benchmark data sets, we carried out a systematic evaluation of ten publicly available AMP prediction methods. Among the six general AMP prediction tools-ADAM, CAMPR3(RF), CAMPR3(SVM), MLAMP, DBAASP and MLAMP-we find that CAMPR3(RF) provides a statistically significant improvement in performance, as measured by the area under the receiver operating characteristic (ROC) curve, relative to the other five methods. Surprisingly, for antibacterial prediction, the original AntiBP method significantly outperforms its successor, AntiBP2 based on one benchmark dataset. The two bacteriocin prediction tools, BAGEL3 and BACTIBASE, both provide very good performance and BAGEL3 outperforms its predecessor, BACTIBASE, on the larger of the two benchmarks. CONTACT: [email protected] or [email protected]. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: Transcription factors (TFs) bind to specific DNA sequence motifs. Several lines of evidence suggest that TF-DNA binding is mediated in part by properties of the local DNA shape: the width of the minor groove, the relative orientations of adjacent base pairs, etc. Several methods have been developed to jointly account for DNA sequence and shape properties in predicting TF binding affinity. However, a limitation of these methods is that they typically require a training set of aligned TF binding sites. RESULTS: We describe a sequence + shape kernel that leverages DNA sequence and shape information to better understand protein-DNA binding preference and affinity. This kernel extends an existing class of k-mer based sequence kernels, based on the recently described di-mismatch kernel. Using three in vitro benchmark datasets, derived from universal protein binding microarrays (uPBMs), genomic context PBMs (gcPBMs) and SELEX-seq data, we demonstrate that incorporating DNA shape information improves our ability to predict protein-DNA binding affinity. In particular, we observe that (i) the k-spectrum + shape model performs better than the classical k-spectrum kernel, particularly for small k values; (ii) the di-mismatch kernel performs better than the k-mer kernel, for larger k; and (iii) the di-mismatch + shape kernel performs better than the di-mismatch kernel for intermediate k values. AVAILABILITY AND IMPLEMENTATION: The software is available at https://bitbucket.org/wenxiu/sequence-shape.git. CONTACT: [email protected] or [email protected]. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

SUMMARY: Genome-wide proximity ligation based assays like Hi-C have opened a window to the 3D organization of the genome. In so doing, they present data structures that are different from conventional 1D signal tracks. To exploit the 2D nature of Hi-C contact maps, matrix techniques like spectral analysis are particularly useful. Here, we present HiC-spector, a collection of matrix-related functions for analyzing Hi-C contact maps. In particular, we introduce a novel reproducibility metric for quantifying the similarity between contact maps based on spectral decomposition. The metric successfully separates contact maps mapped from Hi-C data coming from biological replicates, pseudo-replicates and different cell types. AVAILABILITY AND IMPLEMENTATION: Source code in Julia and Python, and detailed documentation is available at https://github.com/gersteinlab/HiC-spector . CONTACT: [email protected] or [email protected]. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

How should you tackle this task?Ideally, the reviewing process can significantly improve your manuscript by allowing you to take into account the advice of multiple experts in your field.Indeed, empirical evidence suggests that papers that have undergone multiple rounds of peer review fare better in terms of citation counts than papers that are quickly accepted [1].However, in practice, the review process can be emotionally charged as you grapple with comments that may seem to you to be illinformed, biased, or otherwise problematic.A well-crafted "response to reviewers" document is a critical part of your response.This document is submitted alongside your revised manuscript, summarizing the changes that you made in response to the critiques.Too frequently, authors focus on revising the manuscript itself and spend too little time making the response document clear and compelling.The result can be misunderstandings between the reviewers and the authors and ultimately, the possible rejection of a high-quality manuscript.Following are 10 simple rules that can help in formulating an effective response to reviewers. Rule 1: Provide an overview, then quote the full set of reviewsThe response letter will typically begin with a summary of changes, pointing out new data and new analyses performed in response to the most essential criticisms of all the reviewers.Note that, at your discretion, the response may include figures and tables that are for the reviewers' benefit but will not go into the manuscript or supplement.These additional results can be mentioned in your Introduction.If a criticism is raised by multiple reviewers, this can also be pointed out in the summary.Thereafter, the response letter should contain the complete set of reviews with your responses interleaved. Rule 2: Be polite and respectful of all reviewersEven if you are convinced that the reviewer lacks intellectual capacity, it is certainly not in your interest to convey this impression to the reviewer.Keep in mind that if the reviewer failed to understand something, the fault likely lies, at least in part, with you for not making the point clear enough.If the reviewer does not seem to be an expert in the area, remember that this level of expertise (or lack thereof) may be representative of many readers of the journal.Your goal is to make the work clear and accessible to all readers, not just to experts.

UNLABELLED: Precise regulatory control of genes, particularly in eukaryotes, frequently requires the joint action of multiple sequence-specific transcription factors. A cis-regulatory module (CRM) is a genomic locus that is responsible for gene regulation and that contains multiple transcription factor binding sites in close proximity. Given a collection of known transcription factor binding motifs, many bioinformatics methods have been proposed over the past 15 years for identifying within a genomic sequence candidate CRMs consisting of clusters of those motifs. RESULTS: The MCAST algorithm uses a hidden Markov model with a P-value-based scoring scheme to identify candidate CRMs. Here, we introduce a new version of MCAST that offers improved graphical output, a dynamic background model, statistical confidence estimates based on false discovery rate estimation and, most significantly, the ability to predict CRMs while taking into account epigenomic data such as DNase I sensitivity or histone modification data. We demonstrate the validity of MCAST's statistical confidence estimates and the utility of epigenomic priors in identifying CRMs. AVAILABILITY AND IMPLEMENTATION: MCAST is part of the MEME Suite software toolkit. A web server and source code are available at http://meme-suite.org and http://alternate.meme-suite.org CONTACT: [email protected] or [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

UNLABELLED: Tandem mass spectrometry (MS/MS) is the dominant high throughput technology for identifying and quantifying proteins in complex biological samples. Analysis of the tens of thousands of fragmentation spectra produced by an MS/MS experiment begins by assigning to each observed spectrum the peptide that is hypothesized to be responsible for generating the spectrum. This assignment is typically done by searching each spectrum against a database of peptides. To our knowledge, all existing MS/MS search engines compute scores individually between a given observed spectrum and each possible candidate peptide from the database. In this work, we use a trellis, a data structure capable of jointly representing a large set of candidate peptides, to avoid redundantly recomputing common sub-computations among different candidates. We show how trellises may be used to significantly speed up existing scoring algorithms, and we theoretically quantify the expected speedup afforded by trellises. Furthermore, we demonstrate that compact trellis representations of whole sets of peptides enables efficient discriminative learning of a dynamic Bayesian network for spectrum identification, leading to greatly improved spectrum identification accuracy. CONTACT: [email protected] or [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Graph smoothness objectives have achieved great success in semi-supervised learning but have not yet been applied extensively to unsupervised generative models. We define a new class of entropic graph-based posterior regularizers that augment a probabilistic model by encouraging pairs of nearby variables in a regularization graph to have similar posterior distributions. We present a three-way alternating optimization algorithm with closed-form updates for performing inference on this joint model and learning its parameters. This method admits updates linear in the degree of the regularization graph, exhibits monotone convergence and is easily parallelizable. We are motivated by applications in computational biology in which temporal models such as hidden Markov models are used to learn a human-interpretable representation of genomic data. On a synthetic problem, we show that our method outperforms existing methods for graph-based regularization and a comparable strategy for incorporating long-range interactions using existing methods for approximate inference. Using genome-scale functional genomics data, we integrate genome 3D interaction data into existing models for genome annotation and demonstrate significant improvements in predicting genomic activity.

MOTIVATION: Recent technological advances allow the measurement, in a single Hi-C experiment, of the frequencies of physical contacts among pairs of genomic loci at a genome-wide scale. The next challenge is to infer, from the resulting DNA-DNA contact maps, accurate 3D models of how chromosomes fold and fit into the nucleus. Many existing inference methods rely on multidimensional scaling (MDS), in which the pairwise distances of the inferred model are optimized to resemble pairwise distances derived directly from the contact counts. These approaches, however, often optimize a heuristic objective function and require strong assumptions about the biophysics of DNA to transform interaction frequencies to spatial distance, and thereby may lead to incorrect structure reconstruction. METHODS: We propose a novel approach to infer a consensus 3D structure of a genome from Hi-C data. The method incorporates a statistical model of the contact counts, assuming that the counts between two loci follow a Poisson distribution whose intensity decreases with the physical distances between the loci. The method can automatically adjust the transfer function relating the spatial distance to the Poisson intensity and infer a genome structure that best explains the observed data. RESULTS: We compare two variants of our Poisson method, with or without optimization of the transfer function, to four different MDS-based algorithms-two metric MDS methods using different stress functions, a non-metric version of MDS and ChromSDE, a recently described, advanced MDS method-on a wide range of simulated datasets. We demonstrate that the Poisson models reconstruct better structures than all MDS-based methods, particularly at low coverage and high resolution, and we highlight the importance of optimizing the transfer function. On publicly available Hi-C data from mouse embryonic stem cells, we show that the Poisson methods lead to more reproducible structures than MDS-based methods when we use data generated using different restriction enzymes, and when we reconstruct structures at different resolutions. AVAILABILITY AND IMPLEMENTATION: A Python implementation of the proposed method is available at http://cbio.ensmp.fr/pastis.

Cancers arise from successive rounds of mutation and selection, generating clonal populations that vary in size, mutational content and drug responsiveness. Ascertaining the clonal composition of a tumor is therefore important both for prognosis and therapy. Mutation counts and frequencies resulting from next-generation sequencing (NGS) potentially reflect a tumor's clonal composition; however, deconvolving NGS data to infer a tumor's clonal structure presents a major challenge. We propose a generative model for NGS data derived from multiple subsections of a single tumor, and we describe an expectation-maximization procedure for estimating the clonal genotypes and relative frequencies using this model. We demonstrate, via simulation, the validity of the approach, and then use our algorithm to assess the clonal composition of a primary breast cancer and associated metastatic lymph node. After dividing the tumor into subsections, we perform exome sequencing for each subsection to assess mutational content, followed by deep sequencing to precisely count normal and variant alleles within each subsection. By quantifying the frequencies of 17 somatic variants, we demonstrate that our algorithm predicts clonal relationships that are both phylogenetically and spatially plausible. Applying this method to larger numbers of tumors should cast light on the clonal evolution of cancers in space and time.

MOTIVATION: Accurate knowledge of the genome-wide binding of transcription factors in a particular cell type or under a particular condition is necessary for understanding transcriptional regulation. Using epigenetic data such as histone modification and DNase I, accessibility data has been shown to improve motif-based in silico methods for predicting such binding, but this approach has not yet been fully explored. RESULTS: We describe a probabilistic method for combining one or more tracks of epigenetic data with a standard DNA sequence motif model to improve our ability to identify active transcription factor binding sites (TFBSs). We convert each data type into a position-specific probabilistic prior and combine these priors with a traditional probabilistic motif model to compute a log-posterior odds score. Our experiments, using histone modifications H3K4me1, H3K4me3, H3K9ac and H3K27ac, as well as DNase I sensitivity, show conclusively that the log-posterior odds score consistently outperforms a simple binary filter based on the same data. We also show that our approach performs competitively with a more complex method, CENTIPEDE, and suggest that the relative simplicity of the log-posterior odds scoring method makes it an appealing and very general method for identifying functional TFBSs on the basis of DNA and epigenetic evidence. AVAILABILITY AND IMPLEMENTATION: FIMO, part of the MEME Suite software toolkit, now supports log-posterior odds scoring using position-specific priors for motif search. A web server and source code are available at http://meme.nbcr.net. Utilities for creating priors are at http://research.imb.uq.edu.au/t.bailey/SD/Cuellar2011. CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Peptides are routinely identified from mass spectrometry-based proteomics experiments by matching observed spectra to peptides derived from protein databases. The error rates of these identifications can be estimated by target-decoy analysis, which involves matching spectra to shuffled or reversed peptides. Besides estimating error rates, decoy searches can be used by semi-supervised machine learning algorithms to increase the number of confidently identified peptides. As for all machine learning algorithms, however, the results must be validated to avoid issues such as overfitting or biased learning, which would produce unreliable peptide identifications. Here, we discuss how the target-decoy method is employed in machine learning for shotgun proteomics, focusing on how the results can be validated by cross-validation, a frequently used validation scheme in machine learning. We also use simulated data to demonstrate the proposed cross-validation scheme's ability to detect overfitting.

BACKGROUND: Spectral counting methods provide an easy means of identifying proteins with differing abundances between complex mixtures using shotgun proteomics data. The crux spectral-counts command, implemented as part of the Crux software toolkit, implements four previously reported spectral counting methods, the spectral index (SI(N)), the exponentially modified protein abundance index (emPAI), the normalized spectral abundance factor (NSAF), and the distributed normalized spectral abundance factor (dNSAF). RESULTS: We compared the reproducibility and the linearity relative to each protein's abundance of the four spectral counting metrics. Our analysis suggests that NSAF yields the most reproducible counts across technical and biological replicates, and both SI(N) and NSAF achieve the best linearity. CONCLUSIONS: With the crux spectral-counts command, Crux provides open-source modular methods to analyze mass spectrometry data for identifying and now quantifying peptides and proteins. The C++ source code, compiled binaries, spectra and sequence databases are available at http://noble.gs.washington.edu/proj/crux-spectral-counts.

High-throughput proteomics experiments involving tandem mass spectrometry produce large volumes of complex data that require sophisticated computational analyses. As such, the field offers many challenges for computational biologists. In this article, we briefly introduce some of the core computational and statistical problems in the field and then describe a variety of outstanding problems that readers of PLoS Computational Biology might be able to help solve.

The problem of identifying the proteins in a complex mixture using tandem mass spectrometry can be framed as an inference problem on a graph that connects peptides to proteins. Several existing protein identification methods make use of statistical inference methods for graphical models, including expectation maximization, Markov chain Monte Carlo, and full marginalization coupled with approximation heuristics. We show that, for this problem, the majority of the cost of inference usually comes from a few highly connected subgraphs. Furthermore, we evaluate three different statistical inference methods using a common graphical model, and we demonstrate that junction tree inference substantially improves rates of convergence compared to existing methods. The python code used for this paper is available at http://noble.gs.washington.edu/proj/fido.

UNLABELLED: A motif is a short DNA or protein sequence that contributes to the biological function of the sequence in which it resides. Over the past several decades, many computational methods have been described for identifying, characterizing and searching with sequence motifs. Critical to nearly any motif-based sequence analysis pipeline is the ability to scan a sequence database for occurrences of a given motif described by a position-specific frequency matrix. RESULTS: We describe Find Individual Motif Occurrences (FIMO), a software tool for scanning DNA or protein sequences with motifs described as position-specific scoring matrices. The program computes a log-likelihood ratio score for each position in a given sequence database, uses established dynamic programming methods to convert this score to a P-value and then applies false discovery rate analysis to estimate a q-value for each position in the given sequence. FIMO provides output in a variety of formats, including HTML, XML and several Santa Cruz Genome Browser formats. The program is efficient, allowing for the scanning of DNA sequences at a rate of 3.5 Mb/s on a single CPU. AVAILABILITY AND IMPLEMENTATION: FIMO is part of the MEME Suite software toolkit. A web server and source code are available at http://meme.sdsc.edu.

MOTIVATION: A question that often comes up after applying a motif finder to a set of co-regulated DNA sequences is whether the reported putative motif is similar to any known motif. While several tools have been designed for this task, Habib et al. pointed out that the scores that are commonly used for measuring similarity between motifs do not distinguish between a good alignment of two informative columns (say, all-A) and one of two uninformative columns. This observation explains why tools such as Tomtom occasionally return an alignment of uninformative columns which is clearly spurious. To address this problem, Habib et al. suggested a new score [Bayesian Likelihood 2-Component (BLiC)] which uses a Bayesian information criterion to penalize matches that are also similar to the background distribution. RESULTS: We show that the BLiC score exhibits other, highly undesirable properties, and we offer instead a general approach to adjust any motif similarity score so as to reduce the number of reported spurious alignments of uninformative columns. We implement our method in Tomtom and show that, without significantly compromising Tomtom's retrieval accuracy or its runtime, we can drastically reduce the number of uninformative alignments. AVAILABILITY AND IMPLEMENTATION: The modified Tomtom is available as part of the MEME Suite at http://meme.nbcr.net.

BACKGROUND: Protein secondary structure prediction provides insight into protein function and is a valuable preliminary step for predicting the 3D structure of a protein. Dynamic Bayesian networks (DBNs) and support vector machines (SVMs) have been shown to provide state-of-the-art performance in secondary structure prediction. As the size of the protein database grows, it becomes feasible to use a richer model in an effort to capture subtle correlations among the amino acids and the predicted labels. In this context, it is beneficial to derive sparse models that discourage over-fitting and provide biological insight. RESULTS: In this paper, we first show that we are able to obtain accurate secondary structure predictions. Our per-residue accuracy on a well established and difficult benchmark (CB513) is 80.3%, which is comparable to the state-of-the-art evaluated on this dataset. We then introduce an algorithm for sparsifying the parameters of a DBN. Using this algorithm, we can automatically remove up to 70-95% of the parameters of a DBN while maintaining the same level of predictive accuracy on the SD576 set. At 90% sparsity, we are able to compute predictions three times faster than a fully dense model evaluated on the SD576 set. We also demonstrate, using simulated data, that the algorithm is able to recover true sparse structures with high accuracy, and using real data, that the sparse model identifies known correlation structure (local and non-local) related to different classes of secondary structure elements. CONCLUSIONS: We present a secondary structure prediction method that employs dynamic Bayesian networks and support vector machines. We also introduce an algorithm for sparsifying the parameters of the dynamic Bayesian network. The sparsification approach yields a significant speed-up in generating predictions, and we demonstrate that the amino acid correlations identified by the algorithm correspond to several known features of protein secondary structure. Datasets and source code used in this study are available at http://noble.gs.washington.edu/proj/pssp.

BACKGROUND: As genome-wide experiments and annotations become more prevalent, researchers increasingly require tools to help interpret data at this scale. Many functional genomics experiments involve partitioning the genome into labeled segments, such that segments sharing the same label exhibit one or more biochemical or functional traits. For example, a collection of ChlP-seq experiments yields a compendium of peaks, each labeled with one or more associated DNA-binding proteins. Similarly, manually or automatically generated annotations of functional genomic elements, including cis-regulatory modules and protein-coding or RNA genes, can also be summarized as genomic segmentations. RESULTS: We present a software toolkit called Segtools that simplifies and automates the exploration of genomic segmentations. The software operates as a series of interacting tools, each of which provides one mode of summarization. These various tools can be pipelined and summarized in a single HTML page. We describe the Segtools toolkit and demonstrate its use in interpreting a collection of human histone modification data sets and Plasmodium falciparum local chromatin structure data sets. CONCLUSIONS: Segtools provides a convenient, powerful means of interpreting a genomic segmentation.

Virtually every molecular biologist has searched a protein or DNA sequence database to find sequences that are evolutionarily related to a given query. Pairwise sequence comparison methods--i.e., measures of similarity between query and target sequences--provide the engine for sequence database search and have been the subject of 30 years of computational research. For the difficult problem of detecting remote evolutionary relationships between protein sequences, the most successful pairwise comparison methods involve building local models (e.g., profile hidden Markov models) of protein sequences. However, recent work in massive data domains like web search and natural language processing demonstrate the advantage of exploiting the global structure of the data space. Motivated by this work, we present a large-scale algorithm called ProtEmbed, which learns an embedding of protein sequences into a low-dimensional "semantic space." Evolutionarily related proteins are embedded in close proximity, and additional pieces of evidence, such as 3D structural similarity or class labels, can be incorporated into the learning process. We find that ProtEmbed achieves superior accuracy to widely used pairwise sequence methods like PSI-BLAST and HHSearch for remote homology detection; it also outperforms our previous RankProp algorithm, which incorporates global structure in the form of a protein similarity network. Finally, the ProtEmbed embedding space can be visualized, both at the global level and local to a given query, yielding intuition about the structure of protein sequence space.

MOTIVATION: A global map of transcription factor binding sites (TFBSs) is critical to understanding gene regulation and genome function. DNaseI digestion of chromatin coupled with massively parallel sequencing (digital genomic footprinting) enables the identification of protein-binding footprints with high resolution on a genome-wide scale. However, accurately inferring the locations of these footprints remains a challenging computational problem. RESULTS: We present a dynamic Bayesian network-based approach for the identification and assignment of statistical confidence estimates to protein-binding footprints from digital genomic footprinting data. The method, DBFP, allows footprints to be identified in a probabilistic framework and outperforms our previously described algorithm in terms of precision at a fixed recall. Applied to a digital footprinting data set from Saccharomyces cerevisiae, DBFP identifies 4679 statistically significant footprints within intergenic regions. These footprints are mainly located near transcription start sites and are strongly enriched for known TFBSs. Footprints containing no known motif are preferentially located proximal to other footprints, consistent with cooperative binding of these footprints. DBFP also identifies a set of statistically significant footprints in the yeast coding regions. Many of these footprints coincide with the boundaries of antisense transcripts, and the most significant footprints are enriched for binding sites of the chromatin-associated factors Abf1 and Rap1. SUPPLEMENTARY INFORMATION: Supplementary material is available at Bioinformatics online.

SUMMARY: We present a format for efficient storage of multiple tracks of numeric data anchored to a genome. The format allows fast random access to hundreds of gigabytes of data, while retaining a small disk space footprint. We have also developed utilities to load data into this format. We show that retrieving data from this format is more than 2900 times faster than a naive approach using wiggle files. AVAILABILITY AND IMPLEMENTATION: Reference implementation in Python and C components available at http://noble.gs.washington.edu/proj/genomedata/ under the GNU General Public License.

BACKGROUND: Image analysis is an essential component in many biological experiments that study gene expression, cell cycle progression, and protein localization. A protocol for tracking the expression of individual C. elegans genes was developed that collects image samples of a developing embryo by 3-D time lapse microscopy. In this protocol, a program called StarryNite performs the automatic recognition of fluorescently labeled cells and traces their lineage. However, due to the amount of noise present in the data and due to the challenges introduced by increasing number of cells in later stages of development, this program is not error free. In the current version, the error correction (i.e., editing) is performed manually using a graphical interface tool named AceTree, which is specifically developed for this task. For a single experiment, this manual annotation task takes several hours. RESULTS: In this paper, we reduce the time required to correct errors made by StarryNite. We target one of the most frequent error types (movements annotated as divisions) and train a support vector machine (SVM) classifier to decide whether a division call made by StarryNite is correct or not. We show, via cross-validation experiments on several benchmark data sets, that the SVM successfully identifies this type of error significantly. A new version of StarryNite that includes the trained SVM classifier is available at http://starrynite.sourceforge.net. CONCLUSIONS: We demonstrate the utility of a machine learning approach to error annotation for StarryNite. In the process, we also provide some general methodologies for developing and validating a classifier with respect to a given pattern recognition task.

BACKGROUND: The prediction of protein-protein interactions is an important step toward the elucidation of protein functions and the understanding of the molecular mechanisms inside the cell. While experimental methods for identifying these interactions remain costly and often noisy, the increasing quantity of solved 3D protein structures suggests that in silico methods to predict interactions between two protein structures will play an increasingly important role in screening candidate interacting pairs. Approaches using the knowledge of the structure are presumably more accurate than those based on sequence only. Approaches based on docking protein structures solve a variant of this problem, but these methods remain very computationally intensive and will not scale in the near future to the detection of interactions at the level of an interactome, involving millions of candidate pairs of proteins. RESULTS: Here, we describe a computational method to predict efficiently in silico whether two protein structures interact. This yes/no question is presumably easier to answer than the standard protein docking question, "How do these two protein structures interact?" Our approach is to discriminate between interacting and non-interacting protein pairs using a statistical pattern recognition method known as a support vector machine (SVM). We demonstrate that our structure-based method performs well on this task and scales well to the size of an interactome. CONCLUSIONS: The use of structure information for the prediction of protein interaction yields significantly better performance than other sequence-based methods. Among structure-based classifiers, the SVM algorithm, combined with the metric learning pairwise kernel and the MAMMOTH kernel, performs best in our experiments.

Accurately modeling the DNA sequence preferences of transcription factors (TFs), and using these models to predict in vivo genomic binding sites for TFs, are key pieces in deciphering the regulatory code. These efforts have been frustrated by the limited availability and accuracy of TF binding site motifs, usually represented as position-specific scoring matrices (PSSMs), which may match large numbers of sites and produce an unreliable list of target genes. Recently, protein binding microarray (PBM) experiments have emerged as a new source of high resolution data on in vitro TF binding specificities. PBM data has been analyzed either by estimating PSSMs or via rank statistics on probe intensities, so that individual sequence patterns are assigned enrichment scores (E-scores). This representation is informative but unwieldy because every TF is assigned a list of thousands of scored sequence patterns. Meanwhile, high-resolution in vivo TF occupancy data from ChIP-seq experiments is also increasingly available. We have developed a flexible discriminative framework for learning TF binding preferences from high resolution in vitro and in vivo data. We first trained support vector regression (SVR) models on PBM data to learn the mapping from probe sequences to binding intensities. We used a novel -mer based string kernel called the di-mismatch kernel to represent probe sequence similarities. The SVR models are more compact than E-scores, more expressive than PSSMs, and can be readily used to scan genomics regions to predict in vivo occupancy. Using a large data set of yeast and mouse TFs, we found that our SVR models can better predict probe intensity than the E-score method or PBM-derived PSSMs. Moreover, by using SVRs to score yeast, mouse, and human genomic regions, we were better able to predict genomic occupancy as measured by ChIP-chip and ChIP-seq experiments. Finally, we found that by training kernel-based models directly on ChIP-seq data, we greatly improved in vivo occupancy prediction, and by comparing a TF's in vitro and in vivo models, we could identify cofactors and disambiguate direct and indirect binding.

DNA in eukaryotes is packaged into a chromatin complex, the most basic element of which is the nucleosome. The precise positioning of the nucleosome cores allows for selective access to the DNA, and the mechanisms that control this positioning are important pieces of the gene expression puzzle. We describe a large-scale nucleosome pattern that jointly characterizes the nucleosome core and the adjacent linkers and is predominantly characterized by long-range oscillations in the mono, di- and tri-nucleotide content of the DNA sequence, and we show that this pattern can be used to predict nucleosome positions in both Homo sapiens and Saccharomyces cerevisiae more accurately than previously published methods. Surprisingly, in both H. sapiens and S. cerevisiae, the most informative individual features are the mono-nucleotide patterns, although the inclusion of di- and tri-nucleotide features results in improved performance. Our approach combines a much longer pattern than has been previously used to predict nucleosome positioning from sequence-301 base pairs, centered at the position to be scored-with a novel discriminative classification approach that selectively weights the contributions from each of the input features. The resulting scores are relatively insensitive to local AT-content and can be used to accurately discriminate putative dyad positions from adjacent linker regions without requiring an additional dynamic programming step and without the attendant edge effects and assumptions about linker length modeling and overall nucleosome density. Our approach produces the best dyad-linker classification results published to date in H. sapiens, and outperforms two recently published models on a large set of S. cerevisiae nucleosome positions. Our results suggest that in both genomes, a comparable and relatively small fraction of nucleosomes are well-positioned and that these positions are predictable based on sequence alone. We believe that the bulk of the remaining nucleosomes follow a statistical positioning model.

MOTIVATION: A variety of algorithms have been developed to predict transcription factor binding sites (TFBSs) within the genome by exploiting the evolutionary information implicit in multiple alignments of the genomes of related species. One such approach uses an extension of the standard position-specific motif model that incorporates phylogenetic information via a phylogenetic tree and a model of evolution. However, these phylogenetic motif models (PMMs) have never been rigorously benchmarked in order to determine whether they lead to better prediction of TFBSs than obtained using simple position weight matrix scanning. RESULTS: We evaluate three PMM-based prediction algorithms, each of which uses a different treatment of gapped alignments, and we compare their prediction accuracy with that of a non-phylogenetic motif scanning approach. Surprisingly, all of these algorithms appear to be inferior to simple motif scanning, when accuracy is measured using a gold standard of validated yeast TFBSs. However, the PMM scanners perform much better than simple motif scanning when we abandon the gold standard and consider the number of statistically significant sites predicted, using column-shuffled 'random' motifs to measure significance. These results suggest that the common practice of measuring the accuracy of binding site predictors using collections of known sites may be dangerously misleading since such collections may be missing 'weak' sites, which are exactly the type of sites needed to discriminate among predictors. We then extend our previous theoretical model of the statistical power of PMM-based prediction algorithms to allow for loss of binding sites during evolution, and show that it gives a more accurate upper bound on scanner accuracy. Finally, utilizing our theoretical model, we introduce a new method for predicting the number of real binding sites in a genome. The results suggest that the number of true sites for a yeast TF is in general several times greater than the number of known sites listed in the Saccharomyces cerevisiae Database (SCPD). Among the three scanning algorithms that we test, the MONKEY algorithm has the highest accuracy for predicting yeast TFBSs.

UNLABELLED: Qvality is a C++ program for estimating two types of standard statistical confidence measures: the q-value, which is an analog of the p-value that incorporates multiple testing correction, and the posterior error probability (PEP, also known as the local false discovery rate), which corresponds to the probability that a given observation is drawn from the null distribution. In computing q-values, qvality employs a standard bootstrap procedure to estimate the prior probability of a score being from the null distribution; for PEP estimation, qvality relies upon non-parametric logistic regression. Relative to other tools for estimating statistical confidence measures, qvality is unique in its ability to estimate both types of scores directly from a null distribution, without requiring the user to calculate p-values. AVAILABILITY: A web server, C++ source code and binaries are available under MIT license at http://noble.gs.washington.edu/proj/qvality. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

UNLABELLED: We present a large-scale implementation of the Rankprop protein homology ranking algorithm in the form of an openly accessible web server. We use the NRDB40 PSI-BLAST all-versus-all protein similarity network of 1.1 million proteins to construct the graph for the Rankprop algorithm, whereas previously, results were only reported for a database of 108 000 proteins. We also describe two algorithmic improvements to the original algorithm, including propagation from multiple homologs of the query and better normalization of ranking scores, that lead to higher accuracy and to scores with a probabilistic interpretation. AVAILABILITY: The Rankprop web server and source code are available at http://rankprop.gs.washington.edu

Most bioinformatics coursework focuses on algorithms, with perhaps some components devoted to learning programming skills and learning how to use existing bioinformatics software. Unfortunately, for students who are preparing for a research career, this type of curriculum fails to address many of the day-to-day organizational challenges associated with performing computational experiments. In practice, the principles behind organizing and documenting computational experiments are often learned on the fly, and this learning is strongly influenced by personal predilections as well as by chance interactions with collaborators or colleagues. The purpose of this article is to describe one good strategy for carrying out computational experiments. I will not describe profound issues such as how to formulate hypotheses, design experiments, or draw conclusions. Rather, I will focus on relatively mundane issues such as organizing files and directories and documenting progress. These issues are important because poor organizational choices can lead to significantly slower research progress. I do not claim that the strategies I outline here are optimal. These are simply the principles and practices that I have developed over 12 years of bioinformatics research, augmented with various suggestions from other researchers with whom I have discussed these issues.

MOTIVATION: Tandem mass spectrometry (MS/MS) is an indispensable technology for identification of proteins from complex mixtures. Proteins are digested to peptides that are then identified by their fragmentation patterns in the mass spectrometer. Thus, at its core, MS/MS protein identification relies on the relative predictability of peptide fragmentation. Unfortunately, peptide fragmentation is complex and not fully understood, and what is understood is not always exploited by peptide identification algorithms. RESULTS: We use a hybrid dynamic Bayesian network (DBN)/support vector machine (SVM) approach to address these two problems. We train a set of DBNs on high-confidence peptide-spectrum matches. These DBNs, known collectively as Riptide, comprise a probabilistic model of peptide fragmentation chemistry. Examination of the distributions learned by Riptide allows identification of new trends, such as prevalent a-ion fragmentation at peptide cleavage sites C-term to hydrophobic residues. In addition, Riptide can be used to produce likelihood scores that indicate whether a given peptide-spectrum match is correct. A vector of such scores is evaluated by an SVM, which produces a final score to be used in peptide identification. Using Riptide in this way yields improved discrimination when compared to other state-of-the-art MS/MS identification algorithms, increasing the number of positive identifications by as much as 12% at a 1% false discovery rate. AVAILABILITY: Python and C source code are available upon request from the authors. The curated training sets are available at http://noble.gs.washington.edu/proj/intense/. The Graphical Model Tool Kit (GMTK) is freely available at http://ssli.ee.washington.edu/bilmes/gmtk.

MOTIVATION: A recently developed DNaseI assay has given us our first genome-wide view of chromatin structure. In addition to cataloging DNaseI hypersensitive sites, these data allows us to more completely characterize overall features of chromatin accessibility. We employed a Bayesian hierarchical change-point model (CPM), a generalization of a hidden Markov Model (HMM), to characterize tiled microarray DNaseI sensitivity data available from the ENCODE project. RESULTS: Our analysis shows that the accessibility of chromatin to cleavage by DNaseI is well described by a four state model of local segments with each state described by a continuous mixture of Gaussian variables. The CPM produces a better fit to the observed data than the HMM. The large posterior probability for the four-state CPM suggests that the data falls naturally into four classes of regions, which we call major and minor DNaseI hypersensitive sites (DHSs), regions of intermediate sensitivity, and insensitive regions. These classes agree well with a model of chromatin in which local disruptions (DHSs) are concentrated within larger domains of intermediate sensitivity, the accessibility islands. The CPM assigns 92% of the bases within the ENCODE regions to the insensitive regions. The 5.8% of the bases that are in regions of intermediate sensitivity are clearly enriched in functional elements, including genes and activating histone modifications, while the remaining 2.2% of the bases in hypersensitive regions are very strongly enriched in these elements. AVAILABILITY: The CPM software is available upon request from the authors.

BACKGROUND: Predicting a protein's structural or functional class from its amino acid sequence or structure is a fundamental problem in computational biology. Recently, there has been considerable interest in using discriminative learning algorithms, in particular support vector machines (SVMs), for classification of proteins. However, because sufficiently many positive examples are required to train such classifiers, all SVM-based methods are hampered by limited coverage. RESULTS: In this study, we develop a hybrid machine learning approach for classifying proteins, and we apply the method to the problem of assigning proteins to structural categories based on their sequences or their 3D structures. The method combines a full-coverage but lower accuracy nearest neighbor method with higher accuracy but reduced coverage multiclass SVMs to produce a full coverage classifier with overall improved accuracy. The hybrid approach is based on the simple idea of "punting" from one method to another using a learned threshold. CONCLUSION: In cross-validated experiments on the SCOP hierarchy, the hybrid methods consistently outperform the individual component methods at all levels of coverage. Code and data sets are available at http://noble.gs.washington.edu/proj/sabretooth.

Nucleosomes are the fundamental repeating unit of chromatin and comprise the structural building blocks of the living eukaryotic genome. Micrococcal nuclease (MNase) has long been used to delineate nucleosomal organization. Microarray-based nucleosome mapping experiments in yeast chromatin have revealed regularly-spaced translational phasing of nucleosomes. These data have been used to train computational models of sequence-directed nuclesosome positioning, which have identified ubiquitous strong intrinsic nucleosome positioning signals. Here, we successfully apply this approach to nucleosome positioning experiments from human chromatin. The predictions made by the human-trained and yeast-trained models are strongly correlated, suggesting a shared mechanism for sequence-based determination of nucleosome occupancy. In addition, we observed striking complementarity between classifiers trained on experimental data from weakly versus heavily digested MNase samples. In the former case, the resulting model accurately identifies nucleosome-forming sequences; in the latter, the classifier excels at identifying nucleosome-free regions. Using this model we are able to identify several characteristics of nucleosome-forming and nucleosome-disfavoring sequences. First, by combining results from each classifier applied de novo across the human ENCODE regions, the classifier reveals distinct sequence composition and periodicity features of nucleosome-forming and nucleosome-disfavoring sequences. Short runs of dinucleotide repeat appear as a hallmark of nucleosome-disfavoring sequences, while nucleosome-forming sequences contain short periodic runs of GC base pairs. Second, we show that nucleosome phasing is most frequently predicted flanking nucleosome-free regions. The results suggest that the major mechanism of nucleosome positioning in vivo is boundary-event-driven and affirm the classical statistical positioning theory of nucleosome organization.

Proteins do not carry out their functions alone. Instead, they often act by participating in macromolecular complexes and play different functional roles depending on the other members of the complex. It is therefore interesting to identify co-complex relationships. Although protein complexes can be identified in a high-throughput manner by experimental technologies such as affinity purification coupled with mass spectrometry (APMS), these large-scale datasets often suffer from high false positive and false negative rates. Here, we present a computational method that predicts co-complexed protein pair (CCPP) relationships using kernel methods from heterogeneous data sources. We show that a diffusion kernel based on random walks on the full network topology yields good performance in predicting CCPPs from protein interaction networks. In the setting of direct ranking, a diffusion kernel performs much better than the mutual clustering coefficient. In the setting of SVM classifiers, a diffusion kernel performs much better than a linear kernel. We also show that combination of complementary information improves the performance of our CCPP recognizer. A summation of three diffusion kernels based on two-hybrid, APMS, and genetic interaction networks and three sequence kernels achieves better performance than the sequence kernels or diffusion kernels alone. Inclusion of additional features achieves a still better ROC(50) of 0.937. Assuming a negative-to-positive ratio of 600ratio1, the final classifier achieves 89.3% coverage at an estimated false discovery rate of 10%. Finally, we applied our prediction method to two recently described APMS datasets. We find that our predicted positives are highly enriched with CCPPs that are identified by both datasets, suggesting that our method successfully identifies true CCPPs. An SVM classifier trained from heterogeneous data sources provides accurate predictions of CCPPs in yeast. This computational method thereby provides an inexpensive method for identifying protein complexes that extends and complements high-throughput experimental data.

Hidden Markov models (HMMs) have been successfully applied to the tasks of transmembrane protein topology prediction and signal peptide prediction. In this paper we expand upon this work by making use of the more powerful class of dynamic Bayesian networks (DBNs). Our model, Philius, is inspired by a previously published HMM, Phobius, and combines a signal peptide submodel with a transmembrane submodel. We introduce a two-stage DBN decoder that combines the power of posterior decoding with the grammar constraints of Viterbi-style decoding. Philius also provides protein type, segment, and topology confidence metrics to aid in the interpretation of the predictions. We report a relative improvement of 13% over Phobius in full-topology prediction accuracy on transmembrane proteins, and a sensitivity and specificity of 0.96 in detecting signal peptides. We also show that our confidence metrics correlate well with the observed precision. In addition, we have made predictions on all 6.3 million proteins in the Yeast Resource Center (YRC) database. This large-scale study provides an overall picture of the relative numbers of proteins that include a signal-peptide and/or one or more transmembrane segments as well as a valuable resource for the scientific community. All DBNs are implemented using the Graphical Models Toolkit. Source code for the models described here is available at http://noble.gs.washington.edu/proj/philius. A Philius Web server is available at http://www.yeastrc.org/philius, and the predictions on the YRC database are available at http://www.yeastrc.org/pdr.

UNLABELLED: The advent of high-density, high-volume genomic data has created the need for tools to summarize large datasets at multiple scales. HMMSeg is a command-line utility for the scale-specific segmentation of continuous genomic data using hidden Markov models (HMMs). Scale specificity is achieved by an optional wavelet-based smoothing operation. HMMSeg is capable of handling multiple datasets simultaneously, rendering it ideal for integrative analysis of expression, phylogenetic and functional genomic data. AVAILABILITY: http://noble.gs.washington.edu/proj/hmmseg

MOTIVATION: This work aims to develop computational methods to annotate protein structures in an automated fashion. We employ a support vector machine (SVM) classifier to map from a given class of structures to their corresponding structural (SCOP) or functional (Gene Ontology) annotation. In particular, we build upon recent work describing various kernels for protein structures, where a kernel is a similarity function that the classifier uses to compare pairs of structures. RESULTS: We describe a kernel that is derived in a straightforward fashion from an existing structural alignment program, MAMMOTH. We find in our benchmark experiments that this kernel significantly out-performs a variety of other kernels, including several previously described kernels. Furthermore, in both benchmarks, classifying structures using MAMMOTH alone does not work as well as using an SVM with the MAMMOTH kernel. AVAILABILITY: http://noble.gs.washington.edu/proj/3dkernel

The field of computational biology has seen dramatic growth over the past few years, both in terms of available data, scientific questions and challenges for learning and inference. These new types of scientific and clinical problems require the development of novel supervised and unsupervised learning approaches. In particular, the field is characterized by a diversity of heterogeneous data. The human genome sequence is accompanied by real-valued gene expression data, functional annotation of genes, genotyping information, a graph of interacting proteins, a set of equations describing the dynamics of a system, localization of proteins in a cell, a phylogenetic tree relating species, natural language text in the form of papers describing experiments, partial models that provide priors, and numerous other data sources. This supplementary issue consists of seven peer-reviewed papers based on the NIPS Workshop on New Problems and Methods in Computational Biology held at Whistler, British Columbia, Canada on December 8, 2006. The Neural Information Processing Systems Conference is the premier scientific meeting on neural computation, with session topics spanning artificial intelligence, learning theory, neuroscience, etc. The goal of this workshop was to present emerging problems and machine learning techniques in computational biology, with a particular emphasis on methods for computational learning from heterogeneous data. We received 37 extended abstract submissions, from which 13 were selected for oral presentation. The current supplement contains seven papers based on a subset of the 13 extended abstracts. Submitted manuscripts were rigorously reviewed by at least two referees. The quality of each paper was evaluated with respect to its contribution to biology as well as the novelty of the machine learning methods employed.

BACKGROUND: Predicting a protein's structural class from its amino acid sequence is a fundamental problem in computational biology. Much recent work has focused on developing new representations for protein sequences, called string kernels, for use with support vector machine (SVM) classifiers. However, while some of these approaches exhibit state-of-the-art performance at the binary protein classification problem, i.e. discriminating between a particular protein class and all other classes, few of these studies have addressed the real problem of multi-class superfamily or fold recognition. Moreover, there are only limited software tools and systems for SVM-based protein classification available to the bioinformatics community. RESULTS: We present a new multi-class SVM-based protein fold and superfamily recognition system and web server called SVM-Fold, which can be found at http://svm-fold.c2b2.columbia.edu. Our system uses an efficient implementation of a state-of-the-art string kernel for sequence profiles, called the profile kernel, where the underlying feature representation is a histogram of inexact matching k-mer frequencies. We also employ a novel machine learning approach to solve the difficult multi-class problem of classifying a sequence of amino acids into one of many known protein structural classes. Binary one-vs-the-rest SVM classifiers that are trained to recognize individual structural classes yield prediction scores that are not comparable, so that standard "one-vs-all" classification fails to perform well. Moreover, SVMs for classes at different levels of the protein structural hierarchy may make useful predictions, but one-vs-all does not try to combine these multiple predictions. To deal with these problems, our method learns relative weights between one-vs-the-rest classifiers and encodes information about the protein structural hierarchy for multi-class prediction. In large-scale benchmark results based on the SCOP database, our code weighting approach significantly improves on the standard one-vs-all method for both the superfamily and fold prediction in the remote homology setting and on the fold recognition problem. Moreover, our code weight learning algorithm strongly outperforms nearest-neighbor methods based on PSI-BLAST in terms of prediction accuracy on every structure classification problem we consider. CONCLUSION: By combining state-of-the-art SVM kernel methods with a novel multi-class algorithm, the SVM-Fold system delivers efficient and accurate protein fold and superfamily recognition.

BACKGROUND: Much recent work in bioinformatics has focused on the inference of various types of biological networks, representing gene regulation, metabolic processes, protein-protein interactions, etc. A common setting involves inferring network edges in a supervised fashion from a set of high-confidence edges, possibly characterized by multiple, heterogeneous data sets (protein sequence, gene expression, etc.). RESULTS: Here, we distinguish between two modes of inference in this setting: direct inference based upon similarities between nodes joined by an edge, and indirect inference based upon similarities between one pair of nodes and another pair of nodes. We propose a supervised approach for the direct case by translating it into a distance metric learning problem. A relaxation of the resulting convex optimization problem leads to the support vector machine (SVM) algorithm with a particular kernel for pairs, which we call the metric learning pairwise kernel. This new kernel for pairs can easily be used by most SVM implementations to solve problems of supervised classification and inference of pairwise relationships from heterogeneous data. We demonstrate, using several real biological networks and genomic datasets, that this approach often improves upon the state-of-the-art SVM for indirect inference with another pairwise kernel, and that the combination of both kernels always improves upon each individual kernel. CONCLUSION: The metric learning pairwise kernel is a new formulation to infer pairwise relationships with SVM, which provides state-of-the-art results for the inference of several biological networks from heterogeneous genomic data.

The power and popularity of kernel methods stem in part from their ability to handle diverse forms of structured inputs, including vectors, graphs and strings. Recently, several methods have been proposed for combining kernels from heterogeneous data sources. However, all of these methods produce stationary combinations; i.e., the relative weights of the various kernels do not vary among input examples. This article proposes a method for combining multiple kernels in a nonstationary fashion. The approach uses a large-margin latent-variable generative model within the maximum entropy discrimination (MED) framework. Latent parameter estimation is rendered tractable by variational bounds and an iterative optimization procedure. The classifier we use is a log-ratio of Gaussian mixtures, in which each component is implicitly mapped via a Mercer kernel function. We show that the support vector machine is a special case of this model. In this approach, discriminative parameter estimation is feasible via a fast sequential minimal optimization algorithm. Empirical results are presented on synthetic data, several benchmarks, and on a protein function annotation task.

MOTIVATION: Drawing inferences from large, heterogeneous sets of biological data requires a theoretical framework that is capable of representing, e.g. DNA and protein sequences, protein structures, microarray expression data, various types of interaction networks, etc. Recently, a class of algorithms known as kernel methods has emerged as a powerful framework for combining diverse types of data. The support vector machine (SVM) algorithm is the most popular kernel method, due to its theoretical underpinnings and strong empirical performance on a wide variety of classification tasks. Furthermore, several recently described extensions allow the SVM to assign relative weights to various datasets, depending upon their utilities in performing a given classification task. RESULTS: In this work, we empirically investigate the performance of the SVM on the task of inferring gene functional annotations from a combination of protein sequence and structure data. Our results suggest that the SVM is quite robust to noise in the input datasets. Consequently, in the presence of only two types of data, an SVM trained from an unweighted combination of datasets performs as well or better than a more sophisticated algorithm that assigns weights to individual data types. Indeed, for this simple case, we can demonstrate empirically that no solution is significantly better than the naive, unweighted average of the two datasets. On the other hand, when multiple noisy datasets are included in the experiment, then the naive approach fares worse than the weighted approach. Our results suggest that for many applications, a naive unweighted sum of kernels may be sufficient. AVAILABILITY: http://noble.gs.washington.edu/proj/seqstruct

The protein-protein interaction networks of even well-studied model organisms are sketchy at best, highlighting the continued need for computational methods to help direct experimentalists in the search for novel interactions. This need has prompted the development of a number of methods for predicting protein-protein interactions based on various sources of data and methodologies. The common method for choosing negative examples for training a predictor of protein-protein interactions is based on annotations of cellular localization, and the observation that pairs of proteins that have different localization patterns are unlikely to interact. While this method leads to high quality sets of non-interacting proteins, we find that this choice can lead to biased estimates of prediction accuracy, because the constraints placed on the distribution of the negative examples makes the task easier. The effects of this bias are demonstrated in the context of both sequence-based and non-sequence based features used for predicting protein-protein interactions.

BACKGROUND: Biologists regularly search DNA or protein databases for sequences that share an evolutionary or functional relationship with a given query sequence. Traditional search methods, such as BLAST and PSI-BLAST, focus on detecting statistically significant pairwise sequence alignments and often miss more subtle sequence similarity. Recent work in the machine learning community has shown that exploiting the global structure of the network defined by these pairwise similarities can help detect more remote relationships than a purely local measure. METHODS: We review RankProp, a ranking algorithm that exploits the global network structure of similarity relationships among proteins in a database by performing a diffusion operation on a protein similarity network with weighted edges. The original RankProp algorithm is unsupervised. Here, we describe a semi-supervised version of the algorithm that uses labeled examples. Three possible ways of incorporating label information are considered: (i) as a validation set for model selection, (ii) to learn a new network, by choosing which transfer function to use for a given query, and (iii) to estimate edge weights, which measure the probability of inferring structural similarity. RESULTS: Benchmarked on a human-curated database of protein structures, the original RankProp algorithm provides significant improvement over local network search algorithms such as PSI-BLAST. Furthermore, we show here that labeled data can be used to learn a network without any need for estimating parameters of the transfer function, and that diffusion on this learned network produces better results than the original RankProp algorithm with a fixed network. CONCLUSION: In order to gain maximal information from a network, labeled and unlabeled data should be used to extract both local and global structure.

We develop a novel multi-class classification method based on output codes for the problem of classifying a sequence of amino acids into one of many known protein structural classes, called folds. Our method learns relative weights between one-vs-all classifiers and encodes information about the protein structural hierarchy for multi-class prediction. Our code weighting approach significantly improves on the standard one-vs-all method for the fold recognition problem. In order to compare against widely used methods in protein sequence analysis, we also test nearest neighbor approaches based on the PSI-BLAST algorithm. Our code weight learning algorithm strongly outperforms these PSI-BLAST methods on every structure recognition problem we consider. 1.

We describe a hierarchy of motif-based kernels for multiple alignments of biological sequences, particularly suitable to process regulatory regions of genes. The kernels incorporate progressively more information, with the most complex kernel accounting for a multiple alignment of orthologous regions, the phylogenetic tree relating the species, and the prior knowledge that relevant sequence patterns occur in conserved motif blocks. These kernels can be used in the presence of a library of known transcription factor binding sites, or de novo by iterating over all k -mers of a given length. In the latter mode, a discriminative classifier built from such a kernel not only recognizes a given class of promoter regions, but as a side effect simultaneously identifies a collection of relevant, discriminative sequence motifs. We demonstrate the utility of the motif-based multiple alignment kernels by using a collection of aligned promoter regions from five yeast species to recognize classes of cell-cycle regulated genes. Supplementary data is available at http://noble.gs.washington.edu/proj/pkernel.

MOTIVATION: Sequence similarity often suggests evolutionary relationships between protein sequences that can be important for inferring similarity of structure or function. The most widely-used pairwise sequence comparison algorithms for homology detection, such as BLAST and PSI-BLAST, often fail to detect less conserved remotely-related targets. RESULTS: In this paper, we propose a new general graph-based propagation algorithm called MotifProp to detect more subtle similarity relationships than pairwise comparison methods. MotifProp is based on a protein-motif network, in which edges connect proteins and the k-mer based motif features that they contain. We show that our new motif-based propagation algorithm can improve the ranking results over a base algorithm, such as PSI-BLAST, that is used to initialize the ranking. Despite the complex structure of the protein-motif network, MotifProp can be easily interpreted using the top-ranked motifs and motif-rich regions induced by the propagation, both of which are helpful for discovering conserved structural components in remote homologies.

MOTIVATION: Building an accurate protein classification system depends critically upon choosing a good representation of the input sequences of amino acids. Recent work using string kernels for protein data has achieved state-of-the-art classification performance. However, such representations are based only on labeled data--examples with known 3D structures, organized into structural classes--whereas in practice, unlabeled data are far more plentiful. RESULTS: In this work, we develop simple and scalable cluster kernel techniques for incorporating unlabeled data into the representation of protein sequences. We show that our methods greatly improve the classification performance of string kernels and outperform standard approaches for using unlabeled data, such as adding close homologs of the positive examples to the training data. We achieve equal or superior performance to previously presented cluster kernel methods and at the same time achieving far greater computational efficiency. AVAILABILITY: Source code is available at www.kyb.tuebingen.mpg.de/bs/people/weston/semiprot. The Spider matlab package is available at www.kyb.tuebingen.mpg.de/bs/people/spider. SUPPLEMENTARY INFORMATION: www.kyb.tuebingen.mpg.de/bs/people/weston/semiprot.

IN recent years, rapid developments in genomics and proteomics have generated a large amount of data. Often, drawing conclusions from these data requires sophisticated computational analyses. Bioinformatics, or computational biology, is the interdisciplinary science of interpreting biological data using information technology and computer science. The importance of this new field of inquiry will grow as we continue to generate and integrate large quantities of genomic, proteomic, and other data. A particularly active area of research in bioinformatics is the application and development of machine learning techniques to biological problems.Analyzing large biological data sets requires making sense of the data by inferring structure or generalizations from the data. Examples of this type of analysis include protein structure prediction, gene classification, cancer classification based onmicroarray data, clustering of gene expression data, statistical modeling of protein-protein interaction, etc. Each of these tasks can be framed as a problem in machine learning. We therefore see a great potential to increase the interaction between machine learning and bioinformatics. This special issue is aimed at facilitating that interaction. We believe that machine learning can provide powerful tools for analyzing, predicting, and understanding data from emerging genomic and proteomic technologies. The papers submitted to this special issue provide strong evidence that this is the case. In total, more than 50 paperswere submitted to the special issue. After extensive reviews and revisions, 13 papers were accepted into the special issue,whichwill bepublished in two parts. The quality and significance of the accepted papers are very high and the papers cover a wide variety of topics. Below,weprovide a summary of the papers published in this journal issue.

MOTIVATION: During the past decade, the new focus on genomics has highlighted a particular challenge: to integrate the different views of the genome that are provided by various types of experimental data. RESULTS: This paper describes a computational framework for integrating and drawing inferences from a collection of genome-wide measurements. Each dataset is represented via a kernel function, which defines generalized similarity relationships between pairs of entities, such as genes or proteins. The kernel representation is both flexible and efficient, and can be applied to many different types of data. Furthermore, kernel functions derived from different types of data can be combined in a straightforward fashion. Recent advances in the theory of kernel methods have provided efficient algorithms to perform such combinations in a way that minimizes a statistical loss function. These methods exploit semidefinite programming techniques to reduce the problem of finding optimizing kernel combinations to a convex optimization problem. Computational experiments performed using yeast genome-wide datasets, including amino acid sequences, hydropathy profiles, gene expression data and known protein-protein interactions, demonstrate the utility of this approach. A statistical learning algorithm trained from all of these data to recognize particular classes of proteins--membrane proteins and ribosomal proteins--performs significantly better than the same algorithm trained on any single type of data. AVAILABILITY: Supplementary data at http://noble.gs.washington.edu/proj/sdp-svm

MOTIVATION: Classification of proteins sequences into functional and structural families based on sequence homology is a central problem in computational biology. Discriminative supervised machine learning approaches provide good performance, but simplicity and computational efficiency of training and prediction are also important concerns. RESULTS: We introduce a class of string kernels, called mismatch kernels, for use with support vector machines (SVMs) in a discriminative approach to the problem of protein classification and remote homology detection. These kernels measure sequence similarity based on shared occurrences of fixed-length patterns in the data, allowing for mutations between patterns. Thus, the kernels provide a biologically well-motivated way to compare protein sequences without relying on family-based generative models such as hidden Markov models. We compute the kernels efficiently using a mismatch tree data structure, allowing us to calculate the contributions of all patterns occurring in the data in one pass while traversing the tree. When used with an SVM, the kernels enable fast prediction on test sequences. We report experiments on two benchmark SCOP datasets, where we show that the mismatch kernel used with an SVM classifier performs competitively with state-of-the-art methods for homology detection, particularly when very few training examples are available. Examination of the highest-weighted patterns learned by the SVM classifier recovers biologically important motifs in protein families and superfamilies.

Abstract Summary: The support vector machine (SVM) learning algorithm has been widely applied in bioinformatics. We have developed a simple web interface to our implementation of the SVM algorithm, called Gist. This interface allows novice or occasional users to apply a sophisticated machine learning algorithm easily to their data. More advanced users can download the software and source code for local installation. The availability of these tools will permit more widespread application of this powerful learning algorithm in bioinformatics. Availability: Web interface at svm.sdsc.edu. Binaries and source code at microarray.cpmc.columbia.edu/gist.

UNLABELLED: Many types of genomic data can be represented in matrix format, with rows corresponding to genes and columns corresponding to gene features. The heat map is a popular technique for visualizing such data, plotting the data on a two-dimensional grid and using a color scale to represent the magnitude of each matrix entry. Prism is a Web-based software tool for generating annotated heat map visualizations of genome-wide data quickly. The tool provides a selection of genome-specific annotation catalogs as well as a catalog upload capability. The heat maps generated are clickable, allowing the user to drill down to examine specific matrix entries, and gene annotations are linked to relevant genomic databases. AVAILABILITY: http://noble.gs.washington.edu/prism

A key issue in supervised protein classification is the representation of in- put sequences of amino acids. Recent work using string kernels for pro- tein data has achieved state-of-the-art classification performance. How- ever, such representations are based only on labeled data — examples with known 3D structures, organized into structural classes — while in practice, unlabeled data is far more plentiful. In this work, we de- velop simple and scalable cluster kernel techniques for incorporating un- labeled data into the representation of protein sequences. We show that our methods greatly improve the classification performance of string ker- nels and outperform standard approaches for using unlabeled data, such as adding close homologs of the positive examples to the training data. We achieve equal or superior performance to previously presented cluster kernel methods while achieving far greater computational efficiency.

In order to understand the molecular machinery of the cell, we need to know about the multitude of protein-protein interactions that allow the cell to function. High-throughput technologies provide some data about these interactions, but so far that data is fairly noisy. Therefore, computational techniques for predicting protein-protein interactions could be of significant value. One approach to predicting interactions in silico is to produce from first principles a detailed model of a candidate interaction. We take an alternative approach, employing a relatively simple model that learns dynamically from a large collection of data. In this work, we describe an attraction-repulsion model, in which the interaction between a pair of proteins is represented as the sum of attractive and repulsive forces associated with small, domain- or motif-sized features along the length of each protein. The model is discriminative, learning simultaneously from known interactions and from pairs of proteins that are known (or suspected) not to interact. The model is efficient to compute and scales well to very large collections of data. In a cross-validated comparison using known yeast interactions, the attraction-repulsion method performs better than several competing techniques.

MOTIVATION: We examine the effect of replication on the detection of apparently differentially expressed genes in gene expression microarray experiments. Our analysis is based on a random sampling approach using real data sets from 16 published studies. We consider both the ability to find genes that meet particular statistical criteria as well as the stability of the results in the face of changing levels of replication. RESULTS: While dependent on the data source, our findings suggest that stable results are typically not obtained until at least five biological replicates have been used. Conversely, for most studies, 10-15 replicates yield results that are quite stable, and there is less improvement in stability as the number of replicates is further increased. Our methods will be of use in evaluating existing data sets and in helping to design new studies.

UNLABELLED: We describe a simple software tool, 'matrix2png', for creating color images of matrix data. Originally designed with the display of microarray data sets in mind, it is a general tool that can be used to make simple visualizations of matrices for use in figures, web pages, slide presentations and the like. It can also be used to generate images 'on the fly' in web applications. Both continuous-valued and discrete-valued (categorical) data sets can be displayed. Many options are available to the user, including the colors used, the display of row and column labels, and scale bars. In this note we describe some of matrix2png's features and describe some places it has been useful in the authors' work. AVAILABILITY: A simple web interface is available, and Unix binaries are available from http://microarray.cpmc.columbia.edu/matrix2png. Source code is available on request.

MOTIVATION: Unsupervised analysis of microarray gene expression data attempts to find biologically significant patterns within a given collection of expression measurements. For example, hierarchical clustering can be applied to expression profiles of genes across multiple experiments, identifying groups of genes that share similar expression profiles. Previous work using the support vector machine supervised learning algorithm with microarray data suggests that higher-order features, such as pairwise and tertiary correlations across multiple experiments, may provide significant benefit in learning to recognize classes of co-expressed genes. RESULTS: We describe a generalization of the hierarchical clustering algorithm that efficiently incorporates these higher-order features by using a kernel function to map the data into a high-dimensional feature space. We then evaluate the utility of the kernel hierarchical clustering algorithm using both internal and external validation. The experiments demonstrate that the kernel representation itself is insufficient to provide improved clustering performance. We conclude that mapping gene expression data into a high-dimensional feature space is only a good idea when combined with a learning algorithm, such as the support vector machine that does not suffer from the curse of dimensionality. AVAILABILITY: Supplementary data at www.cs.columbia.edu/compbio/hiclust. Software source code available by request.

We introduce a class of string kernels, called mismatch kernels, for use with support vector machines (SVMs) in a discriminative approach to the protein classification problem. These kernels measure sequence sim- ilarity based on shared occurrences of
-length subsequences, counted with up to mismatches, and do not rely on any generative model for the positive training sequences. We compute the kernels efficiently using a mismatch tree data structure and report experiments on a benchmark SCOP dataset, where we show that the mismatch kernel used with an SVM classifier performs as well as the Fisher kernel, the most success- ful method for remote homology detection, while achieving considerable computational savings.

One key element in understanding the molecular machinery of the cell is to understand the meaning, or function, of each protein encoded in the genome. A very successful means of inferring the function of a previously unannotated protein is via sequence similarity with one or more proteins whose functions are already known. Currently, one of the most powerful such homology detection methods is the SVM-Fisher method of Jaakkola, Diekhans and Haussler (ISMB 2000). This method combines a generative, profile hidden Markov model (HMM) with a discriminative classification algorithm known as a support vector machine (SVM). The current work presents an alternative method for SVM-based protein classification. The method, SVM-pairwise, uses a pairwise sequence similarity algorithm such as Smith-Waterman in place of the HMM in the SVM-Fisher method. The resulting algorithm, when tested on its ability to recognize previously unseen families from the SCOP database, yields significantly better remote protein homology detection than SVM-Fisher, profile HMMs and PSI-BLAST.

In our attempts to understand cellular function at the molecular level, we must be able to synthesize information from disparate types of genomic data. We consider the problem of inferring gene functional classifications from a heterogeneous data set consisting of DNA microarray expression measurements and phylogenetic profiles from whole-genome sequence comparisons. We demonstrate the application of the support vector machine (SVM) learning algorithm to this functional inference task. Our results suggest the importance of exploiting prior information about the heterogeneity of the data. In particular, we propose an SVM kernel function that is explicitly heterogeneous. We also show how to use knowledge about heterogeneity to aid in feature selection.

Article Classifying proteins by family using the product of correlated p-values Share on Authors: Timothy L. Bailey NPACI/SDSC, MC 0505, 9500 Gilman Drive, Bldg 109 La Jolla, California NPACI/SDSC, MC 0505, 9500 Gilman Drive, Bldg 109 La Jolla, CaliforniaView Profile , William Noble Grundy Department of Computer Science, University of California, Santa Cruz Santa Cruz, California Department of Computer Science, University of California, Santa Cruz Santa Cruz, CaliforniaView Profile Authors Info & Claims RECOMB '99: Proceedings of the third annual international conference on Computational molecular biologyApril 1999 Pages 10–14https://doi.org/10.1145/299432.299444Published:01 April 1999 10citation359DownloadsMetricsTotal Citations10Total Downloads359Last 12 Months2Last 6 weeks0 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my AlertsNew Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteGet Access

MOTIVATION: Statistical models of protein families, such as position-specific scoring matrices, profiles and hidden Markov models, have been used effectively to find remote homologs when given a set of known protein family members. Unfortunately, training these models typically requires a relatively large set of training sequences. Recent work (Grundy, J. Comput. Biol., 5,<479-492, 1998) has shown that, when only a few family members are known, several theoretically justified statistical modeling techniques fail to provide homology detection performance on a par with Family Pairwise Search (FPS), an algorithm that combines scores from a pairwise sequence similarity algorithm such as BLAST. RESULTS: The present paper provides a model-based algorithm that improves FPS by incorporating hybrid motif-based models of the form generated by Cobbler (Henikoff and Henikoff, Protein Sci., 6, 698-705, 1997). For the 73 protein families investigated here, this cobbled FPS algorithm provides better homology detection performance than either Cobbler or FPS alone. This improvement is maintained when BLAST is replaced with the full Smith-Waterman algorithm. AVAILABILITY: http://fps.sdsc.edu

The function of an unknown biological sequence can often be accurately inferred by identifying sequences homologous to the original sequence. Given a query set of known homologs, there exist at least three general classes of techniques for #nding additional homologs: pairwise sequence comparisons, motif analysis, and hidden Markov modeling. Pairwise sequence comparisons are typically employed when only a single query sequence is known. Hidden Markov models #HMMs#, on the other hand, are usually trained with sets of more than 100 sequences. Motif-based methods fall in between these two extremes. The currentwork compares the performance of representative examples of these three homology detection techniques---using the BLAST, MEME and HMMER software---across a wide range of protein families, using query sets of varying sizes. A linear combination of multiple pairwise sequence comparisons outperforms motif-based and HMM methods for all query set sizes. Furthermore, heuristic pairwise com...

MOTIVATION: Modeling families of related biological sequences using Hidden Markov models (HMMs), although increasingly widespread, faces at least one major problem: because of the complexity of these mathematical models, they require a relatively large training set in order to accurately recognize a given family. For families in which there are few known sequences, a standard linear HMM contains too many parameters to be trained adequately. RESULTS: This work attempts to solve that problem by generating smaller HMMs which precisely model only the conserved regions of the family. These HMMs are constructed from motif models generated by the EM algorithm using the MEME software. Because motif-based HMMs have relatively few parameters, they can be trained using smaller data sets. Studies of short chain alcohol dehydrogenases and 4Fe-4S ferredoxins support the claim that motif-based HMMs exhibit increased sensitivity and selectivity in database searches, especially when training sets contain few sequences.

Many advanced software tools fail to reach a wide audience because they require specialized hardware, installation expertise, or an abundance of CPU cycles. The worldwide web offers a new opportunity for distributing such systems. One such program, MEME, discovers repeated patterns, called motifs, in sets of DNA or protein sequences. This tool is now available to biologists over the worldwide web, using an asynchronous, single-program multiple-data version of the program called ParaMEME that runs on an Intel Paragon XP/S parallel computer at the San Diego Super-computer Center. ParaMEME scales gracefully to 64 nodes on the Paragon with efficiencies > 72% for large data sets. The worldwide web interface to ParaMEME accepts a set of sequences interactively from a user, submits the sequences to the Paragon for analysis, and e-mails the results back to the user. ParaMEME is available for free public use at http://@www.sdsc.edu/CompSci/Biomed/ MEME.

For learning to improve the adaptiveness of an animal's behavior, and thus direct evolution in the way Baldwin suggested, the learning mechanism must incorporate an innate evaluation of how the animal's actions influence its reproductive fitness. For example, many circumstances that damage an animal or otherwise reduce its fitness are painful and tend to be avoided. We refer to the mechanism by which an animal evaluates the fitness consequences of its actions as a “motivation system,” and argue that such a system must evolve along with the behaviors it evaluates. We describe simulations of the evolution of populations of agents instantiating a number of different architectures for generating action and learning in worlds of differing complexity. We find that in some cases, members of the populations evolve motivation systems that are accurate enough to direct learning so as to increase the fitness of the actions that the agents perform. Furthermore, the motivation systems tend to incorporate systematic distortions in their representations of the worlds they inhabit; these distortions can increase the adaptiveness of the behavior generated.