InPrePPI: an integrated evaluation method based on genomic context for predicting protein-protein interactions in prokaryotic genomes

Computational Network Inference for Bacterial Interactomics

Since the large-scale experimental characterization of protein–protein interactions (PPIs) is not possible for all species, several computational PPI prediction methods have been developed that harness existing data from other species. While PPI network prediction has been extensively used in eukaryotes, microbial network inference has lagged behind. However, bacterial interactomes can be built using the same principles and techniques; in fact, several methods are better suited to bacterial genomes. These predicted networks allow systems-level analyses in species that lack experimental interaction data. This review describes the current network inference and analysis techniques and summarizes the use of computationally-predicted microbial interactomes to date.

Prediction of Protein-Protein Interactions by Evidence Combining Methods

Most cellular functions involve proteins' features based on their physical interactions with other partner proteins. Sketching a map of protein-protein interactions (PPIs) is therefore an important inception step towards understanding the basics of cell functions. Several experimental techniques operating in vivo or in vitro have made significant contributions to screening a large number of protein interaction partners, especially high-throughput experimental methods. However, computational approaches for PPI predication supported by rapid accumulation of data generated from experimental techniques, 3D structure definitions, and genome sequencing have boosted the map sketching of PPIs. In this review, we shed light on in silico PPI prediction methods that integrate evidence from multiple sources, including evolutionary relationship, function annotation, sequence/structure features, network topology and text mining. These methods are developed for integration of multi-dimensional evidence, for designing the strategies to predict novel interactions, and for making the results consistent with the increase of prediction coverage and accuracy.

Protein-protein interaction prediction based on multiple kernels and partial network with linear programming

Background

Prediction of de novo protein-protein interaction is a critical step toward reconstructing PPI networks, which is a central task in systems biology. Recent computational approaches have shifted from making PPI prediction based on individual pairs and single data source to leveraging complementary information from multiple heterogeneous data sources and partial network structure. However, how to quickly learn weights for heterogeneous data sources remains a challenge. In this work, we developed a method to infer de novo PPIs by combining multiple data sources represented in kernel format and obtaining optimal weights based on random walk over the existing partial networks.

Results

Our proposed method utilizes Barker algorithm and the training data to construct a transition matrix which constrains how a random walk would traverse the partial network. Multiple heterogeneous features for the proteins in the network are then combined into the form of weighted kernel fusion, which provides a new "adjacency matrix" for the whole network that may consist of disconnected components but is required to comply with the transition matrix on the training subnetwork. This requirement is met by adjusting the weights to minimize the element-wise difference between the transition matrix and the weighted kernels. The minimization problem is solved by linear programming. The weighted kernel fusion is then transformed to regularized Laplacian (RL) kernel to infer missing or new edges in the PPI network, which can potentially connect the previously disconnected components.

Conclusions

The results on synthetic data demonstrated the soundness and robustness of the proposed algorithms under various conditions. And the results on real data show that the accuracies of PPI prediction for yeast data and human data measured as AUC are increased by up to 19 % and 11 % respectively, as compared to a control method without using optimal weights. Moreover, the weights learned by our method Weight Optimization by Linear Programming (WOLP) are very consistent with that learned by sampling, and can provide insights into the relations between PPIs and various feature kernel, thereby improving PPI prediction even for disconnected PPI networks.

Inference of protein-protein interaction networks from multiple heterogeneous data

Protein-protein interaction (PPI) prediction is a central task in achieving a better understanding of cellular and intracellular processes. Because high-throughput experimental methods are both expensive and time-consuming, and are also known of suffering from the problems of incompleteness and noise, many computational methods have been developed, with varied degrees of success. However, the inference of PPI network from multiple heterogeneous data sources remains a great challenge. In this work, we developed a novel method based on approximate Bayesian computation and modified differential evolution sampling (ABC-DEP) and regularized laplacian (RL) kernel. The method enables inference of PPI networks from topological properties and multiple heterogeneous features including gene expression and Pfam domain profiles, in forms of weighted kernels. The optimal weights are obtained by ABC-DEP, and the kernel fusion built based on optimal weights serves as input to RL to infer missing or new edges in the PPI network. Detailed comparisons with control methods have been made, and the results show that the accuracy of PPI prediction measured by AUC is increased by up to 23 %, as compared to a baseline without using optimal weights. The method can provide insights into the relations between PPIs and various feature kernels and demonstrates strong capability of predicting faraway interactions that cannot be well detected by traditional RL method.

Genome-wide protein-protein interactions and protein function exploration in cyanobacteria

Genome-wide network analysis is well implemented to study proteins of unknown function. Here, we effectively explored protein functions and the biological mechanism based on inferred high confident protein-protein interaction (PPI) network in cyanobacteria. We integrated data from seven different sources and predicted 1,997 PPIs, which were evaluated by experiments in molecular mechanism, text mining of literatures in proved direct/indirect evidences, and “interologs” in conservation. Combined the predicted PPIs with known PPIs, we obtained 4,715 no-redundant PPIs (involving 3,231 proteins covering over 90% of genome) to generate the PPI network. Based on the PPI network, terms in Gene ontology (GO) were assigned to function-unknown proteins. Functional modules were identified by dissecting the PPI network into sub-networks and analyzing pathway enrichment, with which we investigated novel function of underlying proteins in protein complexes and pathways. Examples of photosynthesis and DNA repair indicate that the network approach is a powerful tool in protein function analysis. Overall, this systems biology approach provides a new insight into posterior functional analysis of PPIs in cyanobacteria.

Genome-wide prediction of prokaryotic two-component system networks using a sequence-based meta-predictor

BACKGROUND

Two component systems (TCS) are signalling complexes manifested by a histidine kinase (receptor) and a response regulator (effector). They are the most abundant signalling pathways in prokaryotes and control a wide range of biological processes. The pairing of these two components is highly specific, often requiring costly and time-consuming experimental characterisation. Therefore, there is considerable interest in developing accurate prediction tools to lessen the burden of experimental work and cope with the ever-increasing amount of genomic information.

RESULTS

We present a novel meta-predictor, MetaPred2CS, which is based on a support vector machine. MetaPred2CS integrates six sequence-based prediction methods: in-silico two-hybrid, mirror-tree, gene fusion, phylogenetic profiling, gene neighbourhood, and gene operon. To benchmark MetaPred2CS, we also compiled a novel high-quality training dataset of experimentally deduced TCS protein pairs for k-fold cross validation, to act as a gold standard for TCS partnership predictions. Combining individual predictions using MetaPred2CS improved performance when compared to the individual methods and in comparison with a current state-of-the-art meta-predictor.

CONCLUSION

We have developed MetaPred2CS, a support vector machine-based metapredictor for prokaryotic TCS protein pairings. Central to the success of MetaPred2CS is a strategy of integrating individual predictors that improves the overall prediction accuracy, with the in-silico two-hybrid method contributing most to performance. MetaPred2CS outperformed other available systems in our benchmark tests, and is available online at http://metapred2cs.ibers.aber.ac.uk, along with our gold standard dataset of TCS interaction pairs.

Effect of reference genome selection on the performance of computational methods for genome-wide protein-protein interaction prediction

Background

Recent progress in computational methods for predicting physical and functional protein-protein interactions has provided new insights into the complexity of biological processes. Most of these methods assume that functionally interacting proteins are likely to have a shared evolutionary history. This history can be traced out for the protein pairs of a query genome by correlating different evolutionary aspects of their homologs in multiple genomes known as the reference genomes. These methods include phylogenetic profiling, gene neighborhood and co-occurrence of the orthologous protein coding genes in the same cluster or operon. These are collectively known as genomic context methods. On the other hand a method called mirrortree is based on the similarity of phylogenetic trees between two interacting proteins. Comprehensive performance analyses of these methods have been frequently reported in literature. However, very few studies provide insight into the effect of reference genome selection on detection of meaningful protein interactions.

Methods

We analyzed the performance of four methods and their variants to understand the effect of reference genome selection on prediction efficacy. We used six sets of reference genomes, sampled in accordance with phylogenetic diversity and relationship between organisms from 565 bacteria. We used Escherichia coli as a model organism and the gold standard datasets of interacting proteins reported in DIP, EcoCyc and KEGG databases to compare the performance of the prediction methods.

Conclusions

Higher performance for predicting protein-protein interactions was achievable even with 100–150 bacterial genomes out of 565 genomes. Inclusion of archaeal genomes in the reference genome set improves performance. We find that in order to obtain a good performance, it is better to sample few genomes of related genera of prokaryotes from the large number of available genomes. Moreover, such a sampling allows for selecting 50–100 genomes for comparable accuracy of predictions when computational resources are limited.

Do cancer proteins really interact strongly in the human protein-protein interaction network?

Protein-protein interaction (PPI) network analysis has been widely applied in the investigation of the mechanisms of diseases, especially cancer. Recent studies revealed that cancer proteins tend to interact more strongly than other categories of proteins, even essential proteins, in the human interactome. However, it remains unclear whether this observation was introduced by the bias towards more cancer studies in humans. Here, we examined this important issue by uniquely comparing network characteristics of cancer proteins with three other sets of proteins in four organisms, three of which (fly, worm, and yeast) whose interactomes are essentially not biased towards cancer or other diseases. We confirmed that cancer proteins had stronger connectivity, shorter distance, and larger betweenness centrality than non-cancer disease proteins, essential proteins, and control proteins. Our statistical evaluation indicated that such observations were overall unlikely attributed to random events. Considering the large size and high quality of the PPI data in the four organisms, the conclusion that cancer proteins interact strongly in the PPI networks is reliable and robust. This conclusion suggests that perturbation of cancer proteins might cause major changes of cellular systems and result in abnormal cell function leading to cancer.

Modularity analysis based on predicted protein-protein interactions provides new insights into pathogenicity and cellular process of Escherichia coli O157:H7

Background

With the development of experimental techniques and bioinformatics, the quantity of data available from protein-protein interactions (PPIs) is increasing exponentially. Functional modules can be identified from protein interaction networks. It follows that the investigation of functional modules will generate a better understanding of cellular organization, processes, and functions. However, experimental PPI data are still limited, and no modularity analysis of PPIs in pathogens has been published to date.

Results

In this study, we predict and analyze the functional modules of E. coli O157:H7 systemically by integrating several bioinformatics methods. After evaluation, most of the predicted modules are found to be biologically significant and functionally homogeneous. Six pathogenicity-related modules were discovered and analyzed, including novel modules. These modules provided new information on the pathogenicity of O157:H7. The modularity of cellular function and cooperativity between modules are also discussed. Moreover, modularity analysis of O157:H7 can provide possible candidates for biological pathway extension and clues for discovering new pathways of cross-talk.

Conclusions

This article provides the first modularity analysis of a pathogen and sheds new light on the study of pathogens and cellular processes. Our study also provides a strategy for applying modularity analysis to any sequenced organism.

Bayesian inference for genomic data integration reduces misclassification rate in predicting protein-protein interactions

Protein-protein interactions (PPIs) are essential to most fundamental cellular processes. There has been increasing interest in reconstructing PPIs networks. However, several critical difficulties exist in obtaining reliable predictions. Noticeably, false positive rates can be as high as >80%. Error correction from each generating source can be both time-consuming and inefficient due to the difficulty of covering the errors from multiple levels of data processing procedures within a single test. We propose a novel Bayesian integration method, deemed nonparametric Bayes ensemble learning (NBEL), to lower the misclassification rate (both false positives and negatives) through automatically up-weighting data sources that are most informative, while down-weighting less informative and biased sources. Extensive studies indicate that NBEL is significantly more robust than the classic naïve Bayes to unreliable, error-prone and contaminated data. On a large human data set our NBEL approach predicts many more PPIs than naïve Bayes. This suggests that previous studies may have large numbers of not only false positives but also false negatives. The validation on two human PPIs datasets having high quality supports our observations. Our experiments demonstrate that it is feasible to predict high-throughput PPIs computationally with substantially reduced false positives and false negatives. The ability of predicting large numbers of PPIs both reliably and automatically may inspire people to use computational approaches to correct data errors in general, and may speed up PPIs prediction with high quality. Such a reliable prediction may provide a solid platform to other studies such as protein functions prediction and roles of PPIs in disease susceptibility.

Systematic computational prediction of protein interaction networks

Determining the network of physical protein associations is an important first step in developing mechanistic evidence for elucidating biological pathways. Despite rapid advances in the field of high throughput experiments to determine protein interactions, the majority of associations remain unknown. Here we describe computational methods for significantly expanding protein association networks. We describe methods for integrating multiple independent sources of evidence to obtain higher quality predictions and we compare the major publicly available resources available for experimentalists to use.

3D-interologs: an evolution database of physical protein- protein interactions across multiple genomes

Background

Comprehensive exploration of protein-protein interactions is a challenging route to understand biological processes. For efficiently enlarging protein interactions annotated with residue-based binding models, we proposed a new concept "3D-domain interolog mapping" with a scoring system to explore all possible protein pairs between the two homolog families, derived from a known 3D-structure dimmer (template), across multiple species. Each family consists of homologous proteins which have interacting domains of the template for studying domain interface evolution of two interacting homolog families.

Results

The 3D-interologs database records the evolution of protein-protein interactions database across multiple species. Based on "3D-domain interolog mapping" and a new scoring function, we infer 173,294 protein-protein interactions by using 1,895 three-dimensional (3D) structure heterodimers to search the UniProt database (4,826,134 protein sequences). The 3D- interologs database comprises 15,124 species and 283,980 protein-protein interactions, including 173,294 interactions (61%) and 110,686 interactions (39%) summarized from the IntAct database. For a protein-protein interaction, the 3D-interologs database shows functional annotations (e.g. Gene Ontology), interacting domains and binding models (e.g. hydrogen-bond interactions and conserved residues). Additionally, this database provides couple-conserved residues and the interacting evolution by exploring the interologs across multiple species. Experimental results reveal that the proposed scoring function obtains good agreement for the binding affinity of 275 mutated residues from the ASEdb. The precision and recall of our method are 0.52 and 0.34, respectively, by using 563 non-redundant heterodimers to search on the Integr8 database (549 complete genomes).

Conclusions

Experimental results demonstrate that the proposed method can infer reliable physical protein-protein interactions and be useful for studying the protein-protein interaction evolution across multiple species. In addition, the top-ranked strategy and template interface score are able to significantly improve the accuracies of identifying protein-protein interactions in a complete genome. The 3D-interologs database is available at http://3D- interologs.life.nctu.edu.tw.

Global protein interactome exploration through mining genome-scale data in Arabidopsis thaliana

Background

Many essential cellular processes, such as cellular metabolism, transport, cellular metabolism and most regulatory mechanisms, rely on physical interactions between proteins. Genome-wide protein interactome networks of yeast, human and several other animal organisms have already been established, but this kind of network reminds to be established in the field of plant.

Results

We first predicted the protein protein interaction in Arabidopsis thaliana with methods, including ortholog, SSBP, gene fusion, gene neighbor, phylogenetic profile, coexpression, protein domain, and used Naïve Bayesian approach next to integrate the results of these methods and text mining data to build a genome-wide protein interactome network. Furthermore, we adopted the data of GO enrichment analysis, pathway, published literature to validate our network, the confirmation of our network shows the feasibility of using our network to predict protein function and other usage.

Conclusions

Our interactome is a comprehensive genome-wide network in the organism plant Arabidopsis thaliana, and provides a rich resource for researchers in related field to study the protein function, molecular interaction and potential mechanism under different conditions.

A systematic study of genome context methods: calibration, normalization and combination

BACKGROUND

Genome context methods have been introduced in the last decade as automatic methods to predict functional relatedness between genes in a target genome using the patterns of existence and relative locations of the homologs of those genes in a set of reference genomes. Much work has been done in the application of these methods to different bioinformatics tasks, but few papers present a systematic study of the methods and their combination necessary for their optimal use.

RESULTS

We present a thorough study of the four main families of genome context methods found in the literature: phylogenetic profile, gene fusion, gene cluster, and gene neighbor. We find that for most organisms the gene neighbor method outperforms the phylogenetic profile method by as much as 40% in sensitivity, being competitive with the gene cluster method at low sensitivities. Gene fusion is generally the worst performing of the four methods. A thorough exploration of the parameter space for each method is performed and results across different target organisms are presented. We propose the use of normalization procedures as those used on microarray data for the genome context scores. We show that substantial gains can be achieved from the use of a simple normalization technique. In particular, the sensitivity of the phylogenetic profile method is improved by around 25% after normalization, resulting, to our knowledge, on the best-performing phylogenetic profile system in the literature. Finally, we show results from combining the various genome context methods into a single score. When using a cross-validation procedure to train the combiners, with both original and normalized scores as input, a decision tree combiner results in gains of up to 20% with respect to the gene neighbor method. Overall, this represents a gain of around 15% over what can be considered the state of the art in this area: the four original genome context methods combined using a procedure like that used in the STRING database. Unfortunately, we find that these gains disappear when the combiner is trained only with organisms that are phylogenetically distant from the target organism.

CONCLUSIONS

Our experiments indicate that gene neighbor is the best individual genome context method and that gains from the combination of individual methods are very sensitive to the training data used to obtain the combiner's parameters. If adequate training data is not available, using the gene neighbor score by itself instead of a combined score might be the best choice.

Utilizing shared interacting domain patterns and Gene Ontology information to improve protein-protein interaction prediction

Protein-protein interactions (PPIs) play a significant role in many crucial cellular operations such as metabolism, signaling and regulations. The computational methods for predicting PPIs have shown tremendous growth in recent years, but problem such as huge false positive rates has contributed to the lack of solid PPI information. We aimed at enhancing the overlap between computational predictions and experimental results in an effort to partially remove PPIs falsely predicted. The use of protein function predictor named PFP() that are based on shared interacting domain patterns is introduced in this study with the purpose of aiding the Gene Ontology Annotations (GOA). We used GOA and PFP() as agents in a filtering process to reduce false positive pairs in the computationally predicted PPI datasets. The functions predicted by PFP() were extracted from cross-species PPI data in order to assign novel functional annotations for the uncharacterized proteins and also as additional functions for those that are already characterized by the GO (Gene Ontology). The implementation of PFP() managed to increase the chances of finding matching function annotation for the first rule in the filtration process as much as 20%. To assess the capability of the proposed framework in filtering false PPIs, we applied it on the available S. cerevisiae PPIs and measured the performance in two aspects, the improvement made indicated as Signal-to-Noise Ratio (SNR) and the strength of improvement, respectively. The proposed filtering framework significantly achieved better performance than without it in both metrics.

Biomedical discovery acceleration, with applications to craniofacial development

The profusion of high-throughput instruments and the explosion of new results in the scientific literature, particularly in molecular biomedicine, is both a blessing and a curse to the bench researcher. Even knowledgeable and experienced scientists can benefit from computational tools that help navigate this vast and rapidly evolving terrain. In this paper, we describe a novel computational approach to this challenge, a knowledge-based system that combines reading, reasoning, and reporting methods to facilitate analysis of experimental data. Reading methods extract information from external resources, either by parsing structured data or using biomedical language processing to extract information from unstructured data, and track knowledge provenance. Reasoning methods enrich the knowledge that results from reading by, for example, noting two genes that are annotated to the same ontology term or database entry. Reasoning is also used to combine all sources into a knowledge network that represents the integration of all sorts of relationships between a pair of genes, and to calculate a combined reliability score. Reporting methods combine the knowledge network with a congruent network constructed from experimental data and visualize the combined network in a tool that facilitates the knowledge-based analysis of that data. An implementation of this approach, called the Hanalyzer, is demonstrated on a large-scale gene expression array dataset relevant to craniofacial development. The use of the tool was critical in the creation of hypotheses regarding the roles of four genes never previously characterized as involved in craniofacial development; each of these hypotheses was validated by further experimental work.

Gene expression profiling-based in silico approach to identify potential vaccine candidates and drug targets against B. pertussis and B. parapertussis

Whooping cough (pertussis) caused by B. pertussis (B.p) is still serious public health threat. B. parapertussis (B.pp), closely related to B.p, also causes whooping cough. The incidence of B.pp infections has been increasing over the last decades, partly because pertussis vaccines have low efficiency against B.pp infections. Moreover, because the majority of pertussis patients are infants, common antimicrobial agents producing serious adverse reactions in infants are not fully satisfactory. Therefore, we try to identify potential vaccine candidates and alternative drug targets against both B.p and B.pp. This preliminary work integrates several different kinds of data from in silico analysis, comparative genomic hybridization, global transcriptional profiling, and protein-protein interaction (PPI) network to screen potential vaccine candidates and drug targets against the two species. Finally, 191 potential crossprotective vaccine candidates are identified. They have high transcriptional levels in both species, or are associated with virulence and pathogenesis. Moreover, these proteins are not only potentially surface-exposed in the bacteria, but also well conserved among the 165 B.p and B.pp strains. Among them, 22 candidates with high essentiality in the two PPI networks of B.p and B.pp are regarded as suitable drug targets against the two species. We just selected Bordetella as an example to develop a rapid and reliable approach for screening alternative drug targets that associated with novel protein pathways, complexes, and cellular functions against these antibiotic-resistant pathogens. Further researches focusing on the 191 vaccine candidates could accelerate the development of more effective vaccines and drug therapy against B.p and B.pp infection.

Predicting protein linkages in bacteria: which method is best depends on task

Background

Applications of computational methods for predicting protein functional linkages are increasing. In recent years, several bacteria-specific methods for predicting linkages have been developed. The four major genomic context methods are: Gene cluster, Gene neighbor, Rosetta Stone, and Phylogenetic profiles. These methods have been shown to be powerful tools and this paper provides guidelines for when each method is appropriate by exploring different features of each method and potential improvements offered by their combination. We also review many previous treatments of these prediction methods, use the latest available annotations, and offer a number of new observations.

Results

Using Escherichia coli K12 and Bacillus subtilis, linkage predictions made by each of these methods were evaluated against three benchmarks: functional categories defined by COG and KEGG, known pathways listed in EcoCyc, and known operons listed in RegulonDB. Each evaluated method had strengths and weaknesses, with no one method dominating all aspects of predictive ability studied. For functional categories, as previous studies have shown, the Rosetta Stone method was individually best at detecting linkages and predicting functions among proteins with shared KEGG categories while the Phylogenetic profile method was best for linkage detection and function prediction among proteins with common COG functions. Differences in performance under COG versus KEGG may be attributable to the presence of paralogs. Better function prediction was observed when using a weighted combination of linkages based on reliability versus using a simple unweighted union of the linkage sets. For pathway reconstruction, 99 complete metabolic pathways in E. coli K12 (out of the 209 known, non-trivial pathways) and 193 pathways with 50% of their proteins were covered by linkages from at least one method. Gene neighbor was most effective individually on pathway reconstruction, with 48 complete pathways reconstructed. For operon prediction, Gene cluster predicted completely 59% of the known operons in E. coli K12 and 88% (333/418)in B. subtilis. Comparing two versions of the E. coli K12 operon database, many of the unannotated predictions in the earlier version were updated to true predictions in the later version. Using only linkages found by both Gene Cluster and Gene Neighbor improved the precision of operon predictions. Additionally, as previous studies have shown, combining features based on intergenic region and protein function improved the specificity of operon prediction.

Conclusion

A common problem for computational methods is the generation of a large number of false positives that might be caused by an incomplete source of validation. By comparing two versions of a database, we demonstrated the dramatic differences on reported results. We used several benchmarks on which we have shown the comparative effectiveness of each prediction method, as well as provided guidelines as to which method is most appropriate for a given prediction task.