Functional signatures in protein-protein interactions and their impact on signaling pathways.

摘要

Protein-protein interactions (PPIs) are the most fundamental biological processes at the molecular level. PPIs have been proved to be involved in pathologic mechanisms of many diseases. The experimental methods for testing the binding of PPIs are time-consuming and limited by analogs for many reactions. As a result, a computational model is necessary to predict PPIs and to explore the consequences of signal alterations in biological pathways.;A score matrix selection model was built based on overrepresented signature combinations. The case study focused on phosphorylation, which is a well studied post-translational modification category. The signature pairs were extended to signature-string pairs because of the multiple binding sites of kinase/substring interactions. A hypergeometric test was applied to select the significant signals due to the multiple-multiple relationship between the proteins and the domains/motifs. The prediction result shows an extremely high specificity (∼100% compared to random combinations in the human protein pool) and an acceptable sensitivity rate (>65%) according to 10-fold evaluations. The score matrix model has then been extended to the user-defined-input software, named 'YiRen'. A group of PPIs related to transcription factors were evaluated in the test case.;Since the signatures embedded in protein sequences effect signal strength and they could be applied as the predictors in PPIs, alterations of these signatures could lead to broken edges in biological networks. An SNP is a kind of sequence variation. It is the major cause of human genetic variations and plays a key role in personalized medicine. In the DA-SNP (Domain-altering SNP) model, the SNPs from a dbSNP database were filtered through the domain regions on human proteomes. The SNPs were selected if they altered the domain signal strength by more than 10%. Then the selected SNPs were checked through an OMIM database for SNP-disease mappings, while the SNP-corresponding proteins were checked through the protein-disease database in Human Protein Reference Database (HPRD). The altered domains then projected into significant signature vectors in PPI prediction and the broken edges in biological pathways. The model linked the phenotypes and the sequence variation together with functional units in order to provide potential explanations for the phenotypes.