Validity Assessment And Effecting Factors Of Protein-Protein Interaction Prediction Based-on Interologs

Most proteins must be jointed with other proteins to perform its biological functions.Protein-protein interaction (PPI) prediction is one of the most important aspects of the studyon protein functions. PPI networks are valuable for understanding the inner works of mostcomponents of cell, and are meaningful for the study on cell's physiological process. So far,there are only few species or few proteins of some species have their PPI networks been setup because of the limitation of current experiment means. As a result, the amount of PPIsincreases far more behind than that of protein sequences. To accelerate the study on PPI, anumber of new bioinformatics methods were developed, such as phylogenetic profile, geneneighborhood, gene fusion event, correlated mutation, conserved protein-protein interactions(Interologs), and so on. Among these, Interologs is considered briefly and less costly and lesslabor intensive--chinglish. The principle of the Interologs is that physically protein-proteininteractions are evolutionary conserved among organisms. To data, numerous predicted PPIsin some species were based on or partly based on Interologs methods.Although Interologs are widely used to detect PPIs, its validity of detecting PPIs is lack ofevidences. In this article we did an analysis deeply on the validity of predicting PPIs usingInterologs. Using physical PPI information from model organisms such asSaccharomycescerevisiae,Caenorhabditis elegans, Drosophila melanogaster, Homo sapiens,Rattus norvegicus, Mus musculus and Escherichia coli K12, we identified and did a statisticon orthologous proteins among these organisms and examined conserved protein-proteininteractions. Furthermore, we did an analysis on effecting factors of conserved PPIs amongcandidate orthologous proteins. Results indicated that the number of orthologous groupsbetween a pair of species increased with decreasing evolutionary distance between them. Wefound that:1. The Interologs approach with high sensitivity and specificity is efficient to detect PPIs. 2. The approach exhibits the powerful ability to recognize negative interactions frompotential PPIs.3. The conservation of PPI is incorrelated with the connection degrees of proteins in one PPInetwork.4. E value and sequence identity can be used to measure whether the interactions betweenorthologous proteins are conserved.