Explainable convolutional neural network model provides an alternative genome-wide association perspective on mutations in SARS-CoV-2

TL;DR

This study compares an explainable CNN approach with traditional GWAS to identify mutations linked to SARS-CoV-2 variants, demonstrating that neural networks can effectively reveal known mutations and offer a promising alternative for genomic analysis.

Contribution

The paper introduces an explainable CNN model combined with SHAP explanations as a novel method for identifying significant mutations in SARS-CoV-2 genomes, providing an alternative to GWAS.

Findings

CNN outperforms GWAS in identifying known VOC mutations

Explainable AI reveals key nucleotide substitutions in spike gene

Neural network approach offers a promising alternative for genomic analysis

Abstract

Identifying mutations of SARS-CoV-2 strains associated with their phenotypic changes is critical for pandemic prediction and prevention. We compared an explainable convolutional neural network (CNN) approach and the traditional genome-wide association study (GWAS) on the mutations associated with WHO labels of SARS-CoV-2, a proxy for virulence phenotypes. We trained a CNN classification model that can predict genomic sequences into Variants of Concern (VOCs) and then applied Shapley Additive explanations (SHAP) model to identify mutations that are important for the correct predictions. For comparison, we performed traditional GWAS to identify mutations associated with VOCs. Comparison of the two approaches shows that the explainable neural network approach can more effectively reveal known nucleotide substitutions associated with VOCs, such as those in the spike gene regions. Our results suggest that explainable neural networks for genomic sequences offer a promising alternative to the traditional genome wide analysis approaches.