Prediction of drug-target interactions (DTIs) plays a significant role in drug development and drug discovery. Although this task requires a large investment in terms of time and cost, especially when it is performed experimentally, the results are not necessarily significant. Computational DTI prediction is a shortcut to reduce the risks of experimental methods. In this study, we propose an effective approach of nonnegative matrix tri-factorization, referred to as NMTF-DTI, to predict the interaction scores between drugs and targets. NMTF-DTI utilizes multiple kernels (similarity measures) for drugs and targets and Laplacian regularization to boost the prediction performance. The performance of NMTF-DTI is evaluated via cross-validation and is compared with existing DTI prediction methods in terms of the area under the receiver operating characteristic (ROC) curve (AUC) and the area under the precision and recall curve (AUPR). We evaluate our method on four gold standard datasets, comparing to other state-of-the-art methods. Cross-validation and a separate, manually created dataset are used to set parameters. The results show that NMTF-DTI outperforms other competing methods. Moreover, the results of a case study also confirm the superiority of NMTF-DTI.

MOTIVATION: Evidence has shown that microRNAs, one type of small biomolecule, regulate the expression level of genes and play an important role in the development or treatment of diseases. Drugs, as important chemical compounds, can interact with microRNAs and change their functions. The experimental identification of microRNA-drug interactions is time-consuming and expensive. Therefore, it is appealing to develop effective computational approaches for predicting microRNA-drug interactions. RESULTS: In this study, a matrix factorization-based method, called the microRNA-drug interaction prediction approach (MDIPA), is proposed for predicting unknown interactions among microRNAs and drugs. Specifically, MDIPA utilizes experimentally validated interactions between drugs and microRNAs, drug similarity and microRNA similarity to predict undiscovered interactions. A path-based microRNA similarity matrix is constructed, while the structural information of drugs is used to establish a drug similarity matrix. To evaluate its performance, our MDIPA is compared with four state-of-the-art prediction methods with an independent dataset and cross-validation. The results of both evaluation methods confirm the superior performance of MDIPA over other methods. Finally, the results of molecular docking in a case study with breast cancer confirm the efficacy of our approach. In conclusion, MDIPA can be effective in predicting potential microRNA-drug interactions. AVAILABILITY AND IMPLEMENTATION: All code and data are freely available from https://github.com/AliJam82/MDIPA. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

BACKGROUND: Antimicrobial resistance (AMR) is a major threat to global public health because it makes standard treatments ineffective and contributes to the spread of infections. It is important to understand AMR's biological mechanisms for the development of new drugs and more rapid and accurate clinical diagnostics. The increasing availability of whole-genome SNP (single nucleotide polymorphism) information, obtained from whole-genome sequence data, along with AMR profiles provides an opportunity to use feature selection in machine learning to find AMR-associated mutations. This work describes the use of a supervised feature selection approach using deep neural networks to detect AMR-associated genetic factors from whole-genome SNP data. RESULTS: The proposed method, DNP-AAP (deep neural pursuit - average activation potential), was tested on a Neisseria gonorrhoeae dataset with paired whole-genome sequence data and resistance profiles to five commonly used antibiotics including penicillin, tetracycline, azithromycin, ciprofloxacin, and cefixime. The results show that DNP-AAP can effectively identify known AMR-associated genes in N. gonorrhoeae, and also provide a list of candidate genomic features (SNPs) that might lead to the discovery of novel AMR determinants. Logistic regression classifiers were built with the identified SNPs and the prediction AUCs (area under the curve) for penicillin, tetracycline, azithromycin, ciprofloxacin, and cefixime were 0.974, 0.969, 0.949, 0.994, and 0.976, respectively. CONCLUSIONS: DNP-AAP can effectively identify known AMR-associated genes in N. gonorrhoeae. It also provides a list of candidate genes and intergenic regions that might lead to novel AMR factor discovery. More generally, DNP-AAP can be applied to AMR analysis of any bacterial species with genomic variants and phenotype data. It can serve as a useful screening tool for microbiologists to generate genetic candidates for further lab experiments.

De novo peptide sequencing using tandem mass spectrometry (MS/MS) data has become a major computational method for sequence identification in recent years. With the development of new instruments and technology, novel computational methods have emerged with enhanced performance. However, there are only a few methods focusing on ECD/ETD spectra, which mainly contain variants of c -ions and z-ions. Here, a de novo sequencing method for ECD/ETD spectra, NovoExD, is presented. NovoExD applies a new form of spectrum graph with multiple edge types (called a GMET), considers multiple peptide tags, and integrates amino acid combination (AAC) and fragment ion charge information. Its performance is compared with another successful de novo sequencing method, pNovo+, which has an option for ECD/ETD spectra. Experiments conducted on three different datasets show that the average full length peptide identification accuracy of NovoExD is as high as 88.70 percent, and that NovoExD's average accuracy is more than 20 percent greater on all datasets than that of pNovo+.

The majority of scientific resources are devoted to studying a relatively small number of model species, meaning that the ability to translate knowledge across species is of considerable importance. Obtaining species-specific knowledge enables targeted investigations of the biology and pathobiology of a particular species, and facilitates comparative analyses. Phosphorylation is the most widespread posttranslational modification in eukaryotes, and although many phosphorylation sites have been experimentally identified for some species, little or no data are available for others. Using the honeybee as a test organism, this case study illustrates the process of using protein sequence homology to identify putative phosphorylation sites in a species of interest using experimentally determined sites from other species. A number of issues associated with this process are examined and discussed. Several databases of experimentally determined phosphorylation sites exist; however, it can be difficult for the nonspecialist to ascertain how their contents compare. Thus, this case study assesses the content and comparability of several phosphorylation site databases. Additional issues examined include the efficacy of homology-based phosphorylation site prediction, the impact of the level of evolutionary relatedness between species in making these predictions, the ability to translate knowledge of phosphorylation sites across large evolutionary distances and the criteria that should be used in selecting probable phosphorylation sites in the species of interest. Although focusing on phosphorylation, the issues discussed here also apply to the homology-based cross-species prediction of other posttranslational modifications, as well as to sequence motifs in general.

With tandem mass spectrometry (MS/MS), spectra can be generated by various methods including collision-induced dissociation (CID), higher-energy collisional dissociation (HCD), electron capture dissociation (ECD) and electron transfer dissociation (ETD). At the same time, de novo sequencing using multiple spectra from the same peptide is becoming popular in proteomics studies. The focus of this work is using a pair of spectra from CID (or HCD) and ECD (or ETD) fragmentation because of the complementarity between them. We present a new de novo sequencing method for such paired spectra named NovoPair, and compare its performance to another successful method named pNovo+. NovoPair extends our previously proposed graph model to suit paired spectra, and considers different ion types in the two spectra to extract more information. The results show that NovoPair outperforms pNovo+ in terms of full length peptide sequencing accuracy on three pairs of experimental datasets, with the accuracy increasing up to 13.6% compared to pNovo+.

BACKGROUND: There are many programs available for generating simulated whole-genome shotgun sequence reads. The data generated by many of these programs follow predefined models, which limits their use to the authors' original intentions. For example, many models assume that read lengths follow a uniform or normal distribution. Other programs generate models from actual sequencing data, but are limited to reads from single-genome studies. To our knowledge, there are no programs that allow a user to generate simulated data following non-parametric read-length distributions and quality profiles based on empirically-derived information from metagenomics sequencing data. RESULTS: We present BEAR (Better Emulation for Artificial Reads), a program that uses a machine-learning approach to generate reads with lengths and quality values that closely match empirically-derived distributions. BEAR can emulate reads from various sequencing platforms, including Illumina, 454, and Ion Torrent. BEAR requires minimal user input, as it automatically determines appropriate parameter settings from user-supplied data. BEAR also uses a unique method for deriving run-specific error rates, and extracts useful statistics from the metagenomic data itself, such as quality-error models. Many existing simulators are specific to a particular sequencing technology; however, BEAR is not restricted in this way. Because of its flexibility, BEAR is particularly useful for emulating the behaviour of technologies like Ion Torrent, for which no dedicated sequencing simulators are currently available. BEAR is also the first metagenomic sequencing simulator program that automates the process of generating abundances, which can be an arduous task. CONCLUSIONS: BEAR is useful for evaluating data processing tools in genomics. It has many advantages over existing comparable software, such as generating more realistic reads and being independent of sequencing technology, and has features particularly useful for metagenomics work.

BACKGROUND: UniFrac is a well-known tool for comparing microbial communities and assessing statistically significant differences between communities. In this paper we identify a discrepancy in the UniFrac methodology that causes semantically equivalent inputs to produce different outputs in tests of statistical significance. RESULTS: The phylogenetic trees that are input into UniFrac may or may not contain abundance counts. An isomorphic transform can be defined that will convert trees between these two formats without altering the semantic meaning of the trees. UniFrac produces different outputs for these equivalent forms of the same input tree. This is illustrated using metagenomics data from a lake sediment study. CONCLUSIONS: Results from the UniFrac tool can vary greatly for the same input depending on the arbitrary choice of input format. Practitioners should be aware of this issue and use the tool with caution to ensure consistency and validity in their analyses. We provide a script to transform inputs between equivalent formats to help researchers achieve this consistency.

In recent years, de novo peptide sequencing from mass spectrometry data has developed as one of the major peptide identification methods with the emergence of new instruments and advanced computational methods. However, there are still limitations to this method; for example, the typically used spectrum graph model cannot represent all the information and relationships inherent in tandem mass spectra (MS/MS spectra). Here, we present a new spectrum graph model with multiple types of edges (called a multi-edge graph), and integrate amino acid combination (AAC) information and peptide tags into it for peptide sequencing. In addition, the information about immoniun ions observed particularly in higher-energy collisional dissociation (HCD) spectra are incorporated. Comparisons between the proposed method and another successful de novo peptide sequencing method for HCD spectra, pNovo, were performed. Experiments were conducted on four HCD spectral datasets. Results show that the proposed method outperforms pNovo in terms of full length peptide identification accuracy; specifically, the accuracy increases 7%-13% over all four datasets.

SUMMARY: While many experimentally characterized phosphorylation sites exist for certain organisms, such as human, rat and mouse, few sites are known for other organisms, hampering related research efforts. We have developed a software pipeline called DAPPLE that automates the process of using known phosphorylation sites from other organisms to identify putative sites in an organism of interest. AVAILABILITY: DAPPLE is available as a web server at http://saphire.usask.ca. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Abstract Motivation: Phosphorylation is the most important post-translational modification in eukaryotes. Although many computational phosphorylation site prediction tools exist for mammals, and a few were created specifically for Arabidopsis thaliana, none are currently available for other plants. Results: In this article, we propose a novel random forest-based method called PHOSFER (PHOsphorylation Site FindER) for applying phosphorylation data from other organisms to enhance the accuracy of predictions in a target organism. As a test case, PHOSFER is applied to phosphorylation sites in soybean, and we show that it more accurately predicts soybean sites than both the existing Arabidopsis-specific predictors, and a simpler machine-learning scheme that uses only known phosphorylation sites and non-phosphorylation sites from soybean. In addition to soybean, PHOSFER will be extended to other organisms in the near future. Availability: PHOSFER is available via a web interface at http://saphire.usask.ca. Contact: [email protected] Supplementary information: Supplementary data are available at Bioinformatics online.

MOTIVATION: Kinase-mediated phosphorylation is the central mechanism of post-translational modification to regulate cellular responses and phenotypes. Signaling defects associated with protein phosphorylation are linked to many diseases, particularly cancer. Characterizing protein kinases and their substrates enhances our ability to understand and treat such diseases and broadens our knowledge of signaling networks in general. While most or all protein kinases have been identified in well-studied eukaryotes, the sites that they phosphorylate have been only partially elucidated. Experimental methods for identifying phosphorylation sites are resource intensive, so the ability to computationally predict potential sites has considerable value. RESULTS: Many computational techniques for phosphorylation site prediction have been proposed, most of which are available on the web. These techniques differ in several ways, including the machine learning technique used; the amount of sequence information used; whether or not structural information is used in addition to sequence information; whether predictions are made for specific kinases or for kinases in general; and sources of training and testing data. This review summarizes, categorizes and compares the available methods for phosphorylation site prediction, and provides an overview of the challenges that are faced when designing predictors and how they have been addressed. It should therefore be useful both for those wishing to choose a phosphorylation site predictor for their particular biological application, and for those attempting to improve upon established techniques in the future. CONTACT: [email protected].

BACKGROUND: We study the usage of specific peptide platforms in protein composition. Using the pentapeptide as a unit of length, we find that in the universal proteome many pentapeptides are heavily repeated (even thousands of times), whereas some are quite rare, and a small number do not appear at all. To understand the physico-chemical-biological basis underlying peptide usage at the proteomic level, in this study we analyse the energetic costs for the synthesis of rare and never-expressed versus frequent pentapeptides. In addition, we explore residue bulkiness, hydrophobicity, and codon number as factors able to modulate specific peptide frequencies. Then, the possible influence of amino acid composition is investigated in zero- and high-frequency pentapeptide sets by analysing the frequencies of the corresponding inverse-sequence pentapeptides. As a final step, we analyse the pentadecamer oligodeoxynucleotide sequences corresponding to the never-expressed pentapeptides. RESULTS: We find that only DNA context-dependent constraints (such as oligodeoxynucleotide sequence location in the minus strand, introns, pseudogenes, frameshifts, etc.) provide a coherent mechanistic platform to explain the occurrence of never-expressed versus frequent pentapeptides in the protein world. CONCLUSIONS: This study is of importance in cell biology. Indeed, the rarity (or lack of expression) of specific 5-mer peptide modules implies the rarity (or lack of expression) of the corresponding n-mer peptide sequences (with n < 5), so possibly modulating protein compositional trends. Moreover the data might further our understanding of the role exerted by rare pentapeptide modules as critical biological effectors in protein-protein interactions.

BACKGROUND: One type of DNA microarray experiment is discovery of gene expression patterns for a cell line undergoing a biological process over a series of time points. Two important issues with such an experiment are the number of time points, and the interval between them. In the absence of biological knowledge regarding appropriate values, it is natural to question whether the behaviour of progressively generated data may by itself determine a threshold beyond which further microarray experiments do not contribute to pattern discovery. Additionally, such a threshold implies a minimum number of microarray experiments, which is important given the cost of these experiments. RESULTS: We have developed a method for determining the minimum number of microarray experiments (i.e. time points) for temporal gene expression, assuming that the span between time points is given and the hierarchical clustering technique is used for gene expression pattern discovery. The key idea is a similarity measure for two clusterings which is expressed as a function of the data for progressive time points. While the experiments are underway, this function is evaluated. When the function reaches its maximum, it indicates the set of experiments reach a saturated state. Therefore, further experiments do not contribute to the discrimination of patterns. CONCLUSION: The method has been verified with two previously published gene expression datasets. For both experiments, the number of time points determined with our method is less than in the published experiments. It is noted that the overall approach is applicable to other clustering techniques.

Microarray technologies are emerging as a promising tool for genomic studies. A huge body of time-course gene expression data has been and will continuously be produced by microarray experiments. Such gene expression data contains important information and has been proven useful in medical diagnosis, treatment, and drug design. The challenge now is how to analyze such data to obtain the inherent information. Cluster analysis has played an important role in analyzing time-course gene expression data. However, most clustering techniques do not take into consideration the inherent time dependence (dynamics) of time-course gene expression patterns. Accounting for the inherent dynamics of such data in cluster analysis should lead to higher quality clustering. This paper presents a model-based clustering method for time-course gene expression data. The presented method uses Markov chain models (MCMs) to account for the inherent dynamics of time-course gene expression patterns and assumes that expression patterns in the same cluster were generated by the same MCM. For the given number of clusters, the presented method computes cluster models using an EM algorithm and an assignment of genes to these models that maximizes their posterior probabilities. Further, this study employs the average adjusted Rand index (AARI) to evaluate the quality of clustering. The improved performance of the presented method is demonstrated by comparing to the k-means method on a publicly available dataset.

In order to sequence a target DNA, it is first cleaved into many shorter overlapping fragments by chemical or physical techniques. The nucleotide sequence of each fragment is then determined (read) by established methods. The set of all read fragments which cover the target DNA sequence is called its spectrum. It is believed that the shortest superstring of a spectrum is the best candidate for the target DNA sequence. The general problem of finding the shortest superstring for any given set of strings s is NP-hard. Fortunately, the biological instance of this problem is easier. It is not likely that two read fragments, each consisting of several hundred letters, which come from consecutive locations on the target DNA sequence have an overlap of only a few letters; typically, the overlap will be longer. Thus one may reasonably assume that two strings in the spectrum have significant overlap (connectivity) if they come from consecutive locations on the target DNA sequence. A class of important instances satisfying this assumption are those whose spectra are from DNA microarrays. This assumption allows us to claim and show the following: if the spectrum S of a target DNA sequence is substring-free and connected at level t, and the target DNA sequence has no repeats of size t or larger, then there exists an algorithm to reconstruct the target DNA sequence in the linear time O(|S|) after an overlap graph of the spectrum is built.

Gene expression profiles obtained from time-series microarray experiments can reveal important information about biological processes. However, conducting such experiments is costly and time consuming. The cost and time required are linearly proportional to sample size. Therefore, it is worthwhile to provide a way to determine the minimal number of samples or trials required in a microarray experiment. One of the uses of microarray hybridization experiments is to group together genes with similar patterns of the expression using clustering techniques. In this paper, the k-means clustering technique is used. The basic idea of our approach is an incremental process in which testing, analysis and evaluation are integrated and iterated. The process is terminated when the evaluation of the results of two consecutive experiments shows they are sufficiently close. Two measures of "closeness" are proposed and two real microarray datasets are used to validate our approach. The results show that the sample size required to cluster genes in these two datasets can be reduced; i.e. the same results can be achieved with less cost. The approach can be used with other clustering techniques as well.

The hypercube is an interesting and useful topology for parallel computation. While hypercubes have been analyzed in graph theory, this analysis has done little to determine the minimum area required for realizations of the hypercube topology on two-dimensional chips. In a common VLSI layout of the hypercube, the hypercube nodes are placed in a single-row in numeric order. This paper derives an easily computed formula for the minimum number of tracks used by this configuration. For an n-node hypercube, the number of tracks required is roughly two-thirds of n. This result is also a useful upper bound on the number of tracks required under optimal ordering. In general, the number of tracks required is a function of the ordering, but finding the optimal order (optimal in the sense of requiring the minimum number of tracks over all orderings) is NP-hard. Finally, the formula is applied to more area-efficient and practical two-dimensional hypercube layouts. In general, it allows estimation of and control over implementation parameters such as area and chip aspect ratios.

A novel architectural solution for implementing orthogonal persistence in deductive object-oriented database languages is presented. Central to the technique is a persistent type system implemented using distributed objects and ODMG's d Refclass. Use of distributed objects as database types naturally leads to a component-based multi-tier CORBA-like implementation architecture. This architecture seamlessly accommodates object-faulting and transitive persistence. This paper also discusses use of the architecture to implement a deductive object-oriented database system supporting the ROL language and orthogonal persistence.

A multi-structure neural network (MSNN) classifier consisting of four discriminators followed by a maximum selector was designed and applied to classification of four grades of pistachio nuts. Each discriminator was a multi-layer feed-forward neural network with two hidden layers and a single-neuron output layer. Fourier descriptors of the nuts' boundaries and their area were used as the recognition features. The individual discriminators were trained using a biased technique and a back-propagation algorithm. The MSNN classifier gave an average classification performance of 95.0%. This was an increase of 14.8% over the performance of a multi-layer neural network (MLNN) with similar complexity for classifying the same set of patterns.