The practical application of new single molecule protein sequencing (SMPS) technologies requires accurate estimates of their associated sequencing error rates. Here, we describe the development and application of two distinct parameter estimation methods for analyzing SMPS reads produced by fluorosequencing. A Hidden Markov Model (HMM) based approach, extends whatprot, where we previously used HMMs for SMPS peptide-read matching. This extension offers a principled approach for estimating key parameters for fluorosequencing experiments, including missed amino acid cleavages, dye loss, and peptide detachment. Specifically, we adapted the Baum-Welch algorithm, a standard technique to estimate transition probabilities for an HMM using expectation maximization, but modified here to estimate a small number of parameter values directly rather than estimating every transition probability independently. We demonstrate a high degree of accuracy on simulated data, but on experimental datasets, we observed that the model needed to be augmented with an additional error type, N-terminal blocking. This, in combination with data pre-processing, results in reasonable parameterizations of experimental datasets that agree with controlled experimental perturbations. A second independent implementation using a hybrid of DIRECT and Powell's method to reduce the root mean squared error (RMSE) between simulations and the real dataset was also developed. We compare these methods on both simulated and real data, finding that our Baum-Welch based approach outperforms DIRECT and Powell's method by most, but not all, criteria. Although some discrepancies between the results exist, we also find that both approaches provide similar error rate estimates from experimental single molecule fluorosequencing datasets.

BACKGROUND: Proteins often assemble into higher-order complexes to perform their biological functions. Such protein-protein interactions (PPI) are often experimentally measured for pairs of proteins and summarized in a weighted PPI network, to which community detection algorithms can be applied to define the various higher-order protein complexes. Current methods include unsupervised and supervised approaches, often assuming that protein complexes manifest only as dense subgraphs. Utilizing supervised approaches, the focus is not on how to find them in a network, but only on learning which subgraphs correspond to complexes, currently solved using heuristics. However, learning to walk trajectories on a network to identify protein complexes leads naturally to a reinforcement learning (RL) approach, a strategy not extensively explored for community detection. Here, we develop and evaluate a reinforcement learning pipeline for community detection on weighted protein-protein interaction networks to detect new protein complexes. The algorithm is trained to calculate the value of different subgraphs encountered while walking on the network to reconstruct known complexes. A distributed prediction algorithm then scales the RL pipeline to search for novel protein complexes on large PPI networks. RESULTS: The reinforcement learning pipeline is applied to a human PPI network consisting of 8k proteins and 60k PPI, which results in 1,157 protein complexes. The method demonstrated competitive accuracy with improved speed compared to previous algorithms. We highlight protein complexes such as C4orf19, C18orf21, and KIAA1522 which are currently minimally characterized. Additionally, the results suggest TMC04 be a putative additional subunit of the KICSTOR complex and confirm the involvement of C15orf41 in a higher-order complex with HIRA, CDAN1, ASF1A, and by 3D structural modeling. CONCLUSIONS: Reinforcement learning offers several distinct advantages for community detection, including scalability and knowledge of the walk trajectories defining those communities. Applied to currently available human protein interaction networks, this method had comparable accuracy with other algorithms and notable savings in computational time, and in turn, led to clear predictions of protein function and interactions for several uncharacterized human proteins.

We present a machine learning-based interpretive framework (whatprot) for analyzing single molecule protein sequencing data produced by fluorosequencing, a recently developed proteomics technology that determines sparse amino acid sequences for many individual peptide molecules in a highly parallelized fashion. Whatprot uses Hidden Markov Models (HMMs) to represent the states of each peptide undergoing the various chemical processes during fluorosequencing, and applies these in a Bayesian classifier, in combination with pre-filtering by a k-Nearest Neighbors (kNN) classifier trained on large volumes of simulated fluorosequencing data. We have found that by combining the HMM based Bayesian classifier with the kNN pre-filter, we are able to retain the benefits of both, achieving both tractable runtimes and acceptable precision and recall for identifying peptides and their parent proteins from complex mixtures, outperforming the capabilities of either classifier on its own. Whatprot's hybrid kNN-HMM approach enables the efficient interpretation of fluorosequencing data using a full proteome reference database and should now also enable improved sequencing error rate estimates.

The molecular and cellular basis of novelty is an active area of research in evolutionary biology. Until very recently, the vast majority of cellular phenomena were so difficult to sample that cross-species studies of biochemistry were rare and comparative analysis at the level of biochemical systems was almost impossible. Recent advances in systems biology are changing what is possible, however, and comparative phylogenetic methods that can handle this new data are wanted. Here, we introduce the term "phylogenetic latent variable models" (PLVMs, pronounced "plums") for a class of models that has recently been used to infer the evolution of cellular states from systems-level molecular data, and develop a new parameterization and fitting strategy that is useful for comparative inference of biochemical networks. We deploy this new framework to infer the ancestral states and evolutionary dynamics of protein-interaction networks by analyzing >16,000 predominantly metazoan co-fractionation and affinity-purification mass spectrometry experiments. Based on these data, we estimate ancestral interactions across unikonts, broadly recovering protein complexes involved in translation, transcription, proteostasis, transport, and membrane trafficking. Using these results, we predict an ancient core of the Commander complex made up of CCDC22, CCDC93, C16orf62, and DSCR3, with more recent additions of COMMD-containing proteins in tetrapods. We also use simulations to develop model fitting strategies and discuss future model developments.

Determining the three dimensional arrangement of proteins in a complex is highly beneficial for uncovering mechanistic function and interpreting genetic variation in coding genes comprising protein complexes. There are several methods for determining co-complex interactions between proteins, among them co-fractionation / mass spectrometry (CF-MS), but it remains difficult to identify directly contacting subunits within a multi-protein complex. Correlation analysis of CF-MS profiles shows promise in detecting protein complexes as a whole but is limited in its ability to infer direct physical contacts among proteins in sub-complexes. To identify direct protein-protein contacts within human protein complexes we learn a sparse conditional dependency graph from approximately 3,000 CF-MS experiments on human cell lines. We show substantial performance gains in estimating direct interactions compared to correlation analysis on a benchmark of large protein complexes with solved three-dimensional structures. We demonstrate the method's value in determining the three dimensional arrangement of proteins by making predictions for complexes without known structure (the exocyst and tRNA multi-synthetase complex) and by establishing evidence for the structural position of a recently discovered component of the core human EKC/KEOPS complex, GON7/C14ORF142, providing a more complete 3D model of the complex. Direct contact prediction provides easily calculable additional structural information for large-scale protein complex mapping studies and should be broadly applicable across organisms as more CF-MS datasets become available.

MOTIVATION: Novel approaches are needed for discovery of targeted therapies for non-small-cell lung cancer (NSCLC) that are specific to certain patients. Whole genome RNAi screening of lung cancer cell lines provides an ideal source for determining candidate drug targets. RESULTS: Unsupervised learning algorithms uncovered patterns of differential vulnerability across lung cancer cell lines to loss of functionally related genes. Such genetic vulnerabilities represent candidate targets for therapy and are found to be involved in splicing, translation and protein folding. In particular, many NSCLC cell lines were especially sensitive to the loss of components of the LSm2-8 protein complex or the CCT/TRiC chaperonin. Different vulnerabilities were also found for different cell line subgroups. Furthermore, the predicted vulnerability of a single adenocarcinoma cell line to loss of the Wnt pathway was experimentally validated with screening of small-molecule Wnt inhibitors against an extensive cell line panel. AVAILABILITY AND IMPLEMENTATION: The clustering algorithm is implemented in Python and is freely available at https://bitbucket.org/youngjh/nsclc_paper CONTACT: [email protected] or [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

How do bacteria regulate their cellular physiology in response to starvation? Here, we present a detailed characterization of Escherichia coli growth and starvation over a time-course lasting two weeks. We have measured multiple cellular components, including RNA and proteins at deep genomic coverage, as well as lipid modifications and flux through central metabolism. Our study focuses on the physiological response of E. coli in stationary phase as a result of being starved for glucose, not on the genetic adaptation of E. coli to utilize alternative nutrients. In our analysis, we have taken advantage of the temporal correlations within and among RNA and protein abundances to identify systematic trends in gene regulation. Specifically, we have developed a general computational strategy for classifying expression-profile time courses into distinct categories in an unbiased manner. We have also developed, from dynamic models of gene expression, a framework to characterize protein degradation patterns based on the observed temporal relationships between mRNA and protein abundances. By comparing and contrasting our transcriptomic and proteomic data, we have identified several broad physiological trends in the E. coli starvation response. Strikingly, mRNAs are widely down-regulated in response to glucose starvation, presumably as a strategy for reducing new protein synthesis. By contrast, protein abundances display more varied responses. The abundances of many proteins involved in energy-intensive processes mirror the corresponding mRNA profiles while proteins involved in nutrient metabolism remain abundant even though their corresponding mRNAs are down-regulated.

The proteomes of cells, tissues, and organisms reflect active cellular processes and change continuously in response to intracellular and extracellular cues. Deep, quantitative profiling of the proteome, especially if combined with mRNA and metabolite measurements, should provide an unprecedented view of cell state, better revealing functions and interactions of cell components. Molecular diagnostics and biomarker discovery should benefit particularly from the accurate quantification of proteomes, since complex diseases like cancer change protein abundances and modifications. Currently, shotgun mass spectrometry is the primary technology for high-throughput protein identification and quantification; while powerful, it lacks high sensitivity and coverage. We draw parallels with next-generation DNA sequencing and propose a strategy, termed fluorosequencing, for sequencing peptides in a complex protein sample at the level of single molecules. In the proposed approach, millions of individual fluorescently labeled peptides are visualized in parallel, monitoring changing patterns of fluorescence intensity as N-terminal amino acids are sequentially removed, and using the resulting fluorescence signatures (fluorosequences) to uniquely identify individual peptides. We introduce a theoretical foundation for fluorosequencing and, by using Monte Carlo computer simulations, we explore its feasibility, anticipate the most likely experimental errors, quantify their potential impact, and discuss the broad potential utility offered by a high-throughput peptide sequencing technology.

BACKGROUND: Phenotypes and diseases may be related to seemingly dissimilar phenotypes in other species by means of the orthology of underlying genes. Such "orthologous phenotypes," or "phenologs," are examples of deep homology, and may be used to predict additional candidate disease genes. RESULTS: In this work, we develop an unsupervised algorithm for ranking phenolog-based candidate disease genes through the integration of predictions from the k nearest neighbor phenologs, comparing classifiers and weighting functions by cross-validation. We also improve upon the original method by extending the theory to paralogous phenotypes. Our algorithm makes use of additional phenotype data--from chicken, zebrafish, and E. coli, as well as new datasets for C. elegans--establishing that several types of annotations may be treated as phenotypes. We demonstrate the use of our algorithm to predict novel candidate genes for human atrial fibrillation (such as HRH2, ATP4A, ATP4B, and HOPX) and epilepsy (e.g., PAX6 and NKX2-1). We suggest gene candidates for pharmacologically-induced seizures in mouse, solely based on orthologous phenotypes from E. coli. We also explore the prediction of plant gene-phenotype associations, as for the Arabidopsis response to vernalization phenotype. CONCLUSIONS: We are able to rank gene predictions for a significant portion of the diseases in the Online Mendelian Inheritance in Man database. Additionally, our method suggests candidate genes for mammalian seizures based only on bacterial phenotypes and gene orthology. We demonstrate that phenotype information may come from diverse sources, including drug sensitivities, gene ontology biological processes, and in situ hybridization annotations. Finally, we offer testable candidates for a variety of human diseases, plant traits, and other classes of phenotypes across a wide array of species.

MOTIVATION: A number of computational methods have been proposed that predict protein-protein interactions (PPIs) based on protein sequence features. Since the number of potential non-interacting protein pairs (negative PPIs) is very high both in absolute terms and in comparison to that of interacting protein pairs (positive PPIs), computational prediction methods rely upon subsets of negative PPIs for training and validation. Hence, the need arises for subset sampling for negative PPIs. RESULTS: We clarify that there are two fundamentally different types of subset sampling for negative PPIs. One is subset sampling for cross-validated testing, where one desires unbiased subsets so that predictive performance estimated with them can be safely assumed to generalize to the population level. The other is subset sampling for training, where one desires the subsets that best train predictive algorithms, even if these subsets are biased. We show that confusion between these two fundamentally different types of subset sampling led one study recently published in Bioinformatics to the erroneous conclusion that predictive algorithms based on protein sequence features are hardly better than random in predicting PPIs. Rather, both protein sequence features and the 'hubbiness' of interacting proteins contribute to effective prediction of PPIs. We provide guidance for appropriate use of random versus balanced sampling. AVAILABILITY: The datasets used for this study are available at http://www.marcottelab.org/PPINegativeDataSampling. CONTACT: [email protected]; [email protected]. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: High-throughput protein identification experiments based on tandem mass spectrometry (MS/MS) often suffer from low sensitivity and low-confidence protein identifications. In a typical shotgun proteomics experiment, it is assumed that all proteins are equally likely to be present. However, there is often other evidence to suggest that a protein is present and confidence in individual protein identification can be updated accordingly. RESULTS: We develop a method that analyzes MS/MS experiments in the larger context of the biological processes active in a cell. Our method, MSNet, improves protein identification in shotgun proteomics experiments by considering information on functional associations from a gene functional network. MSNet substantially increases the number of proteins identified in the sample at a given error rate. We identify 8-29% more proteins than the original MS experiment when applied to yeast grown in different experimental conditions analyzed on different MS/MS instruments, and 37% more proteins in a human sample. We validate up to 94% of our identifications in yeast by presence in ground-truth reference sets. AVAILABILITY AND IMPLEMENTATION: Software and datasets are available at http://aug.csres.utexas.edu/msnet

MOTIVATION: Tandem mass spectrometry (MS/MS) offers fast and reliable characterization of complex protein mixtures, but suffers from low sensitivity in protein identification. In a typical shotgun proteomics experiment, it is assumed that all proteins are equally likely to be present. However, there is often other information available, e.g. the probability of a protein's presence is likely to correlate with its mRNA concentration. RESULTS: We develop a Bayesian score that estimates the posterior probability of a protein's presence in the sample given its identification in an MS/MS experiment and its mRNA concentration measured under similar experimental conditions. Our method, MSpresso, substantially increases the number of proteins identified in an MS/MS experiment at the same error rate, e.g. in yeast, MSpresso increases the number of proteins identified by approximately 40%. We apply MSpresso to data from different MS/MS instruments, experimental conditions and organisms (Escherichia coli, human), and predict 19-63% more proteins across the different datasets. MSpresso demonstrates that incorporating prior knowledge of protein presence into shotgun proteomics experiments can substantially improve protein identification scores. AVAILABILITY AND IMPLEMENTATION: Software is available upon request from the authors. Mass spectrometry datasets and supplementary information are available from (http://www.marcottelab.org/MSpresso/).

UNLABELLED: Mass spectrometry-based proteomics stands to gain from additional analysis of its data, but its large, complex datasets make demands on speed and memory usage requiring special consideration from scripting languages. The software library 'mspire'-developed in the Ruby programming language-offers quick and memory-efficient readers for standard xml proteomics formats, converters for intermediate file types in typical proteomics spectral-identification work flows (including the Bioworks .srf format), and modules for the calculation of peptide false identification rates. AVAILABILITY: Freely available at http://mspire.rubyforge.org. Additional data models, usage information, and methods available at http://bioinformatics.icmb.utexas.edu/mspire

BACKGROUND: Mass spectrometry (MS) based label-free protein quantitation has mainly focused on analysis of ion peak heights and peptide spectral counts. Most analyses of tandem mass spectrometry (MS/MS) data begin with an enzymatic digestion of a complex protein mixture to generate smaller peptides that can be separated and identified by an MS/MS instrument. Peptide spectral counting techniques attempt to quantify protein abundance by counting the number of detected tryptic peptides and their corresponding MS spectra. However, spectral counting is confounded by the fact that peptide physicochemical properties severely affect MS detection resulting in each peptide having a different detection probability. Lu et al. (2007) described a modified spectral counting technique, Absolute Protein Expression (APEX), which improves on basic spectral counting methods by including a correction factor for each protein (called Oi value) that accounts for variable peptide detection by MS techniques. The technique uses machine learning classification to derive peptide detection probabilities that are used to predict the number of tryptic peptides expected to be detected for one molecule of a particular protein (Oi). This predicted spectral count is compared to the protein's observed MS total spectral count during APEX computation of protein abundances. RESULTS: The APEX Quantitative Proteomics Tool, introduced here, is a free open source Java application that supports the APEX protein quantitation technique. The APEX tool uses data from standard tandem mass spectrometry proteomics experiments and provides computational support for APEX protein abundance quantitation through a set of graphical user interfaces that partition th parameter controls for the various processing tasks. The tool also provides a Z-score analysis for identification of significant differential protein expression, a utility to assess APEX classifier performance via cross validation, and a utility to merge multiple APEX results into a standardized format in preparation for further statistical analysis. CONCLUSION: The APEX Quantitative Proteomics Tool provides a simple means to quickly derive hundreds to thousands of protein abundance values from standard liquid chromatography-tandem mass spectrometry proteomics datasets. The APEX tool provides a straightforward intuitive interface design overlaying a highly customizable computational workflow to produce protein abundance values from LC-MS/MS datasets.

Proteins interact in complex protein-protein interaction (PPI) networks whose topological properties-such as scale-free topology, hierarchical modularity, and dissortativity-have suggested models of network evolution. Currently preferred models invoke preferential attachment or gene duplication and divergence to produce networks whose topology matches that observed for real PPIs, thus supporting these as likely models for network evolution. Here, we show that the interaction density and homodimeric frequency are highly protein age-dependent in real PPI networks in a manner which does not agree with these canonical models. In light of these results, we propose an alternative stochastic model, which adds each protein sequentially to a growing network in a manner analogous to protein crystal growth (CG) in solution. The key ideas are (1) interaction probability increases with availability of unoccupied interaction surface, thus following an anti-preferential attachment rule, (2) as a network grows, highly connected sub-networks emerge into protein modules or complexes, and (3) once a new protein is committed to a module, further connections tend to be localized within that module. The CG model produces PPI networks consistent in both topology and age distributions with real PPI networks and is well supported by the spatial arrangement of protein complexes of known 3-D structure, suggesting a plausible physical mechanism for network evolution.

BACKGROUND: Identifying all protein complexes in an organism is a major goal of systems biology. In the past 18 months, the results of two genome-scale tandem affinity purification-mass spectrometry (TAP-MS) assays in yeast have been published, along with corresponding complex maps. For most complexes, the published data sets were surprisingly uncorrelated. It is therefore useful to consider the raw data from each study and generate an accurate complex map from a high-confidence data set that integrates the results of these and earlier assays. RESULTS: Using an unsupervised probabilistic scoring scheme, we assigned a confidence score to each interaction in the matrix-model interpretation of the large-scale yeast mass-spectrometry data sets. The scoring metric proved more accurate than the filtering schemes used in the original data sets. We then took a high-confidence subset of these interactions and derived a set of complexes using MCL. The complexes show high correlation with existing annotations. Hierarchical organization of some protein complexes is evident from inter-complex interactions. CONCLUSION: We demonstrate that our scoring method can generate an integrated high-confidence subset of observed matrix-model interactions, which we subsequently used to derive an accurate map of yeast complexes. Our results indicate that essentiality is a product of the protein complex rather than the individual protein, and that we have achieved near saturation of the yeast high-abundance, rich-media-expressed "complex-ome."

MOTIVATION: We reformulate the problem of comparing mass-spectra by mapping spectra to a vector space model. Our search method leverages a metric space indexing algorithm to produce an initial candidate set, which can be followed by any fine ranking scheme. RESULTS: We consider three distance measures integrated into a multi-vantage point index structure. Of these, a semi-metric fuzzy-cosine distance using peptide precursor mass constraints performs the best. The index acts as a coarse, lossless filter with respect to the SEQUEST and ProFound scoring schemes, reducing the number of distance computations and returned candidates for fine filtering to about 0.5% and 0.02% of the database respectively. The fuzzy cosine distance term improves specificity over a peptide precursor mass filter, reducing the number of returned candidates by an order of magnitude. Run time measurements suggest proportional speedups in overall search times. Using an implementation of ProFound's Bayesian score as an example of a fine filter on a test set of Escherichia coli protein fragmentation spectra, the top results of our sample system are consistent with that of SEQUEST.

UNLABELLED: We introduce the Protein Link EXplorer (PLEX), a web-based environment that allows the construction of a phylogenetic profile for any given amino acid sequence, and its comparison with profiles of approximately 350,000 predicted genes from 89 genomes, as a means of interactively identifying functionally linked genes and predicting protein function. PLEX can be searched iteratively and also enables searches for chromosomal gene neighbors and Rosetta Stone linkages. PLEX search results are accompanied by quantitative estimates of linkage confidence, enabling users to take advantage of coinheritance, operon and gene fusion-based methods for inferring gene function and reconstructing cellular systems and pathways. AVAILABILITY: http://bioinformatics.icmb.utexas.edu/plex

MOTIVATION: Clustering genes based upon their expression patterns allows us to predict gene function. Most existing clustering algorithms cluster genes together when their expression patterns show high positive correlation. However, it has been observed that genes whose expression patterns are strongly anti-correlated can also be functionally similar. Biologically, this is not unintuitive-genes responding to the same stimuli, regardless of the nature of the response, are more likely to operate in the same pathways. RESULTS: We present a new diametrical clustering algorithm that explicitly identifies anti-correlated clusters of genes. Our algorithm proceeds by iteratively (i). re-partitioning the genes and (ii). computing the dominant singular vector of each gene cluster; each singular vector serving as the prototype of a 'diametric' cluster. We empirically show the effectiveness of the algorithm in identifying diametrical or anti-correlated clusters. Testing the algorithm on yeast cell cycle data, fibroblast gene expression data, and DNA microarray data from yeast mutants reveals that opposed cellular pathways can be discovered with this method. We present systems whose mRNA expression patterns, and likely their functions, oppose the yeast ribosome and proteosome, along with evidence for the inverse transcriptional regulation of a number of cellular systems.

The amount of data produced by molecular biologists is growing at an exponential rate. Some of the fastest growing sets of data are measurements of gene expression, comparable in quantity only to gene sequences and the vast biological literature. Both gene expression data and sequence data offer hints as to the functions of thousands of newly discovered genes, but neither give complete answers. Therefore, much effort is being focused on integrating these large data sets and combining them with all available functional data to draw inferences about the functions of uncharacterised genes. This review discusses the most pertinent functional data for genome-wide functional inference and describes several methods by which these disparate data types are being integrated.

Abstract Motivation: A central problem in bioinformatics is how to capture information from the vast current scientific literature in a form suitable for analysis by computer. We address the special case of information on protein–protein interactions, and show that the frequencies of words in Medline abstracts can be used to determine whether or not a given paper discusses protein–protein interactions. For those papers determined to discuss this topic, the relevant information can be captured for the Database of Interacting Proteins. Furthermore, suitable gene annotations can also be captured. Results: Our Bayesian approach scores Medline abstracts for probability of discussing the topic of interest according to the frequencies of discriminating words found in the abstract. More than 80 discriminating words (e.g. complex, interaction, two-hybrid) were determined from a training set of 260 Medline abstracts corresponding to previously validated entries in the Database of Interacting Proteins. Using these words and a log likelihood scoring function, \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \({\sim}\) \end{document}2000 Medline abstracts were identified as describing interactions between yeast proteins. This approach now forms the basis for the rapid expansion of the Database of Interacting Proteins. Contact: [email protected]; [email protected]; [email protected] * These authors contributed equally to this work.