Evaluation of physical and functional protein-protein interaction prediction methods for detecting biological pathways

Computational Methods and Deep Learning for Elucidating Protein Interaction Networks

Protein interactions play a critical role in all biological processes, but experimental identification of protein interactions is a time- and resource-intensive process. The advances in next-generation sequencing and multi-omics technologies have greatly benefited large-scale predictions of protein interactions using machine learning methods. A wide range of tools have been developed to predict protein-protein, protein-nucleic acid, and protein-drug interactions. Here, we discuss the applications, methods, and challenges faced when employing the various prediction methods. We also briefly describe ways to overcome the challenges and prospective future developments in the field of protein interaction biology.

Search, Retrieve, Visualize, and Analyze Protein-Protein Interactions from Multiple Databases: A Guide for Experimental Biologists

Functional annotation is lacking for over half of the proteins encoded in genomes and model or representative organisms are not an exception to this trend. One of the popular ways of assigning putative functions to uncharacterized proteins is based on the functions of well-characterized proteins that physically interact with them, i.e., guilt-by-association or functional context approach. In the last two decades, several powerful experimental and computational techniques have been used to determine protein-protein interactions (PPIs) at genome level and are made available through many public databases. The PPI data are often complex and heterogeneously represented across databases posing unique challenges in retrieving, integrating, and analyzing the data even for trained computational biologists, the end users-experimental biologists often struggle to work around the data for the protein of their interests. This chapter provides stepwise protocols to import interaction network of the protein of interest in Cytoscape using PSICQUIC, stringApp, and IntAct App. These are next-generation applications that import PPI from multiple databases/resources and provide seamless functions to study the protein of interest and its functional context directly in Cytoscape.

Human transcriptional gene regulatory network compiled from 14 data resources

The transcriptional regulatory network (TRN) in a cell orchestrates spatio-temporal expression of genes to generate cellular responses for maintenance, reproduction, development and survival of the cell and its hosting organism. Transcription factors (TF) regulate the expression of their target genes (TG) and are the fundamental units of TRN. Several databases have been developed to catalogue human TRN based on low- and high-throughput experimental and computational studies considering their importance in understanding cellular physiology. However, literature lacks their comparative assessment on the strengths and weaknesses. We compared over 2.2 million regulatory pairs between 1379 TF and 22,518 TG from 14 resources. Our study reveals that the TF and TG were common across data resources but not their regulatory pairs. TF and TG of the regulatory pairs showed weak expression correlation, significant gene ontology overlap, co-citations in PubMed and low numbers of TF-TG pairs representing transcriptional repression relationships. We assigned each TF-TG regulatory pair a combined confidence score reflecting its reliability based on its presence in multiple databases. The assembled TRN contains 2,246,598 TF-TG pairs, of which, 44,284 with information on TF's activating or repressing effects on their TG and is available upon request. This study brings the information about transcriptional regulation scattered over the literature and databases at one place in the form of one of the most comprehensive and complete human TRN assembled to date. It will be a valuable resource for benchmarking TRN prediction tools, and to the scientific community working in functional genomics, gene expression and regulation analysis.

Inter-protein residue covariation information unravels physically interacting protein dimers

BACKGROUND

Predicting physical interaction between proteins is one of the greatest challenges in computational biology. There are considerable various protein interactions and a huge number of protein sequences and synthetic peptides with unknown interacting counterparts. Most of co-evolutionary methods discover a combination of physical interplays and functional associations. However, there are only a handful of approaches which specifically infer physical interactions. Hybrid co-evolutionary methods exploit inter-protein residue coevolution to unravel specific physical interacting proteins. In this study, we introduce a hybrid co-evolutionary-based approach to predict physical interplays between pairs of protein families, starting from protein sequences only.

RESULTS

In the present analysis, pairs of multiple sequence alignments are constructed for each dimer and the covariation between residues in those pairs are calculated by CCMpred (Contacts from Correlated Mutations predicted) and three mutual information based approaches for ten accessible surface area threshold groups. Then, whole residue couplings between proteins of each dimer are unified into a single Frobenius norm value. Norms of residue contact matrices of all dimers in different accessible surface area thresholds are fed into support vector machine as single or multiple feature models. The results of training the classifiers by single features show no apparent different accuracies in distinct methods for different accessible surface area thresholds. Nevertheless, mutual information product and context likelihood of relatedness procedures may roughly have an overall higher and lower performances than other two methods for different accessible surface area cut-offs, respectively. The results also demonstrate that training support vector machine with multiple norm features for several accessible surface area thresholds leads to a considerable improvement of prediction performance. In this context, CCMpred roughly achieves an overall better performance than mutual information based approaches. The best accuracy, sensitivity, specificity, precision and negative predictive value for that method are 0.98, 1, 0.962, 0.96, and 0.962, respectively.

CONCLUSIONS

In this paper, by feeding norm values of protein dimers into support vector machines in different accessible surface area thresholds, we demonstrate that even small number of proteins in pairs of multiple alignments could allow one to accurately discriminate between positive and negative dimers.

PDZ Domains Across the Microbial World: Molecular Link to the Proteases, Stress Response, and Protein Synthesis

The PSD-95/Dlg-A/ZO-1 (PDZ) domain is highly expanded, diversified, and well distributed across metazoa where it assembles diverse signaling components by virtue of interactions with other proteins in a sequence-specific manner. In contrast, in the microbial world they are reported to be involved in protein quality control during stress response. The distribution, functions, and origins of PDZ domain-containing proteins in the prokaryotic organisms remain largely unexplored. We analyzed 7,852 PDZ domain-containing proteins in 1,474 microbial genomes in this context. PDZ domain-containing proteins from planctomycetes, myxobacteria, and other eubacteria occupying terrestrial and aquatic niches are found to be in multiple copies within their genomes. Over 93% of the 7,852 PDZ domain-containing proteins were classified into 12 families including six novel families based on additional structural and functional domains present in these proteins. The higher PDZ domain encoding capacity of the investigated organisms was observed to be associated with adaptation to the ecological niche where multicellular life might have originated and flourished. Predicted subcellular localization of PDZ domain-containing proteins and their genomic context argue in favor of crucial roles in translation and membrane remodeling during stress response. Based on rigorous sequence, structure, and phylogenetic analyses, we propose that the highly diverse PDZ domain of the uncharacterized Fe-S oxidoreductase superfamily, exclusively found in gladobacteria and several anaerobes and acetogens, might represent the most ancient form among all the existing PDZ domains.

Structural bioinformatics of the interactome

The past decade has seen a dramatic expansion in the number and range of techniques available to obtain genome-wide information and to analyze this information so as to infer both the functions of individual molecules and how they interact to modulate the behavior of biological systems. Here, we review these techniques, focusing on the construction of physical protein-protein interaction networks, and highlighting approaches that incorporate protein structure, which is becoming an increasingly important component of systems-level computational techniques. We also discuss how network analyses are being applied to enhance our basic understanding of biological systems and their disregulation, as well as how these networks are being used in drug development.