SCORPION is a stacking-based ensemble learning framework for accurate prediction of phage virion proteins

Saeed Ahmad,Mohammad Ali Moni,Watshara Shoombuatong,Phasit Charoenkwan,Pietro Lio’,Md Mehedi Hasan,Julian M W Quinn

doi:10.1038/s41598-022-08173-5

Abstract

Fast and accurate identification of phage virion proteins (PVPs) would greatly aid facilitation of antibacterial drug discovery and development. Although, several research efforts based on machine learning (ML) methods have been made for in silico identification of PVPs, these methods have certain limitations. Therefore, in this study, we propose a new computational approach, termed SCORPION, (StaCking-based Predictior fOR Phage VIrion PrOteiNs), to accurately identify PVPs using only protein primary sequences. Specifically, we explored comprehensive 13 different feature descriptors from different aspects (i.e., compositional information, composition-transition-distribution information, position-specific information and physicochemical properties) with 10 popular ML algorithms to construct a pool of optimal baseline models. These optimal baseline models were then used to generate probabilistic features (PFs) and considered as a new feature vector. Finally, we utilized a two-step feature selection strategy to determine the optimal PF feature vector and used this feature vector to develop a stacked model (SCORPION). Both tenfold cross-validation and independent test results indicate that SCORPION achieves superior predictive performance than its constitute baseline models and existing methods. We anticipate SCORPION will serve as a useful tool for the cost-effective and large-scale screening of new PVPs. The source codes and datasets for this work are available for downloading in the GitHub repository (https://github.com/saeed344/SCORPION).

Highlights

Fast and accurate identification of phage virion proteins (PVPs) would greatly aid facilitation of antibacterial drug discovery and development
We introduced SCORPION, a novel, stacked, machine learning-based approach for accurate identification of PVPs
These baseline models were used to generate and construct the probabilistic features (PFs) feature vector, which were considered as new feature representations

Summary

Introduction

Fast and accurate identification of phage virion proteins (PVPs) would greatly aid facilitation of antibacterial drug discovery and development. We explored comprehensive 13 different feature descriptors from different aspects (i.e., compositional information, composition-transition-distribution information, position-specific information and physicochemical properties) with 10 popular ML algorithms to construct a pool of optimal baseline models These optimal baseline models were used to generate probabilistic features (PFs) and considered as a new feature vector. Current experimental approaches for the identification of PVPs from non-PVPs include many scientific instruments and methodologies such as mass spectrometry, sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-GE) based proteomic methods and protein analysis arrays[5–7].

Methods

Results

Conclusion