PPI_SVM: Prediction of protein-protein interactions using machine learning, domain-domain affinities and frequency tables

Abstract

Protein-protein interactions (PPI) control most of the biological processes in a living cell. In order to fully understand protein functions, a knowledge of protein-protein interactions is necessary. Prediction of PPI is challenging, especially when the three-dimensional structure of interacting partners is not known. Recently, a novel prediction method was proposed by exploiting physical interactions of constituent domains. We propose here a novel knowledge-based prediction method, namely PPI_SVM, which predicts interactions between two protein sequences by exploiting their domain information. We trained a two-class support vector machine on the benchmarking set of pairs of interacting proteins extracted from the Database of Interacting Proteins (DIP). The method considers all possible combinations of constituent domains between two protein sequences, unlike most of the existing approaches. Moreover, it deals with both single-domain proteins and multiUnauthenticated Download Date | 5/13/18 1:16 AM CELLULAR & MOLECULAR BIOLOGY LETTERS 265 domain proteins; therefore it can be applied to the whole proteome in highthroughput studies. Our machine learning classifier, following a brainstorming approach, achieves accuracy of 86%, with specificity of 95%, and sensitivity of 75%, which are better results than most previous methods that sacrifice recall values in order to boost the overall precision. Our method has on average better sensitivity combined with good selectivity on the benchmarking dataset. The PPI_SVM source code, train/test datasets and supplementary files are available freely in the public domain

FunPred-1: Protein function prediction from a protein interaction network using neighborhood analysis

“…For performance improvements, domain-domain affinity information may be incorporated in prediction of the protein functions. The PPIs may be decomposed into physical interactions between constituent domains of proteins, using the method proposed in one of our earlier works [19]. The use of domain interaction information in the prediction of protein function may be considered as a future extension of this study.…”

Protein features identification for machine learning-based prediction of protein-protein interactions

“…They are used to predict potential interactions between proteins, to validate results of highthroughput interaction screens and to analyze the protein networks inferred from interaction databases. Several statistical and machine learning based methods have been applied for the prediction of PPI including Bayesian Networks (Jansen et al, 2003;Patil & Nakamura, 2005), Simple Naïve Bayesian, Random Forest (Šikic et al 2009;Zubek et al, 2015), Support Vector Machine (Bock & Gough, 2001;Chatterjee et al, 2011;You et al, 2013;You et al, 2014;Zubek et al, 2015), Decision Tree, Logistic Regression, k-Nearest Neighborhood (kNN), Conditional Random Field, Artificial Neural Networks (Fariselli et al, 2002), to name a few. Despite the success of these methods, there is still need for the improvement in terms of prediction accuracy and computational efficiency (Res et al, 2005, Bordner & Abagyan, 2005.…”

Review of computational methods for virus–host protein interaction prediction: a case study on novel Ebola–human interactions

“…The computational PPI prediction methods primarily use sequence information [9][10][11][12][13][14], domain-based [5,[15][16][17], protein structure [18][19][20][21], physiochemical properties [22], semantic analysis [23,24] and known interactions between virus and host proteins [25]. Classical machine learning techniques are well-accepted tools for PPI predictions when there are sufficient numbers of known interactions.…”