Comprehensive Feature Extraction Model to Classify Interacting and Non-Interacting Proteins in Human Viruses using Random Forest Classifier

Abstract

Abstract Protein-protein interactions are crucial for the entry of viruses into the cell. Understanding the mechanism of interactions is essential in studying human-virus association, developing new biologics and drug candidates, as well as viral infections and antiviral responses. Experimental methods to analyze human-virus protein-protein interactions are time-consuming and labor-intensive, so machine learning based methods are being developed to predict interactions and determine large-scale interactomes between species. The present work highlights the importance of features in the classification of interacting and non-interacting proteins. To achieve this, we have extracted all possible features like Amino Acid Composition (AAC), Dipeptides Composition (DPC), Grouped Amino Acid Composition (GAAC), Pseudo-Amino Acid Composition (PAAC) etc. that can be fetched from a protein sequence which lies in a higher dimension space. We have used a random forest classifier to understand the biological relevance of these high-dimensional features and thereafter to decide whether these features really contribute to the protein-protein interactions. As part of this, the classifier was applied to three datasets, two with dimensionality reduction and one without dimensionality reduction. The datasets in which dimensionality reduction is applied produce 100% accuracy and one without dimensionality reduction gave 85% accuracy. Thus it is evident that dimensionality reduction fails to capture the complexity of biological relevance and the underlying associations between human and viral proteins.