Knowledge-guided inference of domain–domain interactions from incomplete protein–protein interaction networks

Abstract

Motivation: Protein-protein interactions (PPIs), though extremely valuable towards a better understanding of protein functions and cellular processes, do not provide any direct information about the regions/domains within the proteins that mediate the interaction. Most often, it is only a fraction of a protein that directly interacts with its biological partners. Thus, understanding interaction at the domain level is a critical step towards (i) thorough understanding of PPI networks; (ii) precise identification of binding sites; (iii) acquisition of insights into the causes of deleterious mutations at interaction sites; and (iv) most importantly, development of drugs to inhibit pathological protein interactions. In addition, knowledge derived from known domain–domain interactions (DDIs) can be used to understand binding interfaces, which in turn can help discover unknown PPIs.Results: Here, we describe a novel method called K-GIDDI (knowledge-guided inference of DDIs) to narrow down the PPI sites to smaller regions/domains. K-GIDDI constructs an initial DDI network from cross-species PPI networks, and then expands the DDI network by inferring additional DDIs using a divide-and-conquer biclustering algorithm guided by Gene Ontology (GO) information, which identifies partial-complete bipartite sub-networks in the DDI network and makes them complete bipartite sub-networks by adding edges. Our results indicate that K-GIDDI can reliably predict DDIs. Most importantly, K-GIDDI's novel network expansion procedure allows prediction of DDIs that are otherwise not identifiable by methods that rely only on PPI data.Contact: xwchen@ku.eduAvailability: http://www.ittc.ku.edu/∼xwchen/domainNetwork/ddinet.htmlSupplementary information: Supplementary data are available at Bioinformatics online.

MYBBP1A: a new Ipr1’s binding protein in mice

“…In order to identify more potential canine Ipr1 binding partners, we discarded those with repeated gene names. Since two domains are more likely to have similar GO terms if they are interacting [35], each candidate proteins' GO annotation terms were checked based on the Ipr1's cellular location and its role in biological processes. Those with a GO annotation term of nuclear and terms of transcription/regulation of transcription/apoptosis or child terms of these annotations were kept.…”

Reconstituting protein interaction networks using parameter-dependent domain-domain interactions

“…Existing computational methods use varying approaches to tackle different aspects of these problems, each with its own set of aims, strengths, and limitations [18-30]. For example, some methods use additional biological information, such as gene expression data, to establish whether a PPI can occur [24,25,29], and others limit the PPI coverage to smaller sets of high-confidence interactions [19,26,27]. An additional promising approach is to use a feature selection algorithm to find a set of DDIs that best discriminate between true and false PPIs [30].…”

Predicting conformation of protein complexes by determining statistically significant domain-domain interactions

“…In the context of protein interaction, such domains and peptides act as recognition elements; we refer to these binding domains or recognized peptides simply as 'domains' in this study. Over the past few years with developments of highthroughput PPI detection technologies, many researchers have shown an interest in extracting domain-domain interactions (DDIs) from large-scale PPI data by statistical methods, demonstrating that the idea of DDIs explain the cause of PPIs in some measure (Sprinzak and Margalit 2001;Riley et al 2005;Singhal and Resat 2007;Liu et al 2009). Here, we statistically extracted DDIs from integrated PPI data of Arabidopsis by following a procedure described in the 'Method' section.…”

“…These studies have reported that the densely connected regions in a network correspond to known protein complexes or protein functional units. The other approach is the analysis of DDIs coming out from PPI data by statistics or machine learning intended to predict unknown PPIs (Sprinzak and Margalit 2001;Riley et al 2005;Singhal and Resat 2007;Liu et al 2009). These studies have shown that the concept of DDIs statistically extracted from large-scale PPI data can explain the makeup of PPIs to some extent.…”