Pred-AHCP: Robust feature selection enabled Sequence Specific Prediction of Anti-Hepatitis C Peptides via Machine Learning

Abstract

Every year, an estimated 1.5 million people worldwide contract Hepatitis C (HepC), a significant contributor to liver disease. Although many studies have explored machine learnings potential to predict antiviral peptides, very few have addressed predicting peptides against specific viruses such as Hepatitis C. In this study, we demonstrate the use of machine learning (ML) algorithms to predict peptides that are effective against HepC. We developed an explainable ML model that harnesses the amino acid sequence of a peptide to predict its potential as an anti-HepC (AHC) agent. Specifically, features were computed based on sequence and physicochemical properties, with feature selection performed utilizing a combined scheme of mutual information and variance inflation factor. This facilitated the removal of redundant and multicollinear features from the sequence data, enhancing the models generalizability in predicting AHCPs. The model using the random forest algorithm produced the best performance with an accuracy of about 90%. The feature selection analysis highlights that the distribution of hydrophobicity and polarizability, as well as the frequencies of glycine residues and di-peptide motifs--YXL, LXK, VXXXF, VL, LV, CC, RR, TXXXV, VXXA, CXXXC--emerged as the key predictors for identifying AHCPs targeting different components of the HepC virus. The model developed can be accessed through the Pred-AHCP web server, provided at http://tinyurl.com/web-Pred-AHCP. This resource facilitates the prediction and re-engineering of AHCPs for designing peptide-based therapeutics while also proposing an exploration of similar strategies for designing peptide inhibitors effective against other viruses.