Explainable machine learning reveals evolutionary signals in influenza haemagglutinin.

Abstract

Identifying amino acid changes that lead to phenotypic change is critical to viral surveillance. Common metrics, such as site-wise evolutionary rates and entropy, are true measures of variability rather than phenotypic importance. Here, I show that supervised, explainable machine learning (ML) models provide a complementary approach to date and classify sequences, identify mutations for host adaptation and control for confounding covariates. I used 39 121 haemagglutinin (H3) protein sequences with passage annotations to create models of sequence change. Gradient-boosted decision trees were trained with encoded amino acids; SHapley Additive exPlanations (SHAP) values quantified site importance. The passage classifier achieved 81% overall accuracy, distinguishing egg grown from unpassaged isolates with nearly 90% recall. A separate regressor predicted sample collection date with R2=0.98 and a mean absolute error of 74.5 days. Moreover, the sites identified as most important showed a strong enrichment for experimentally validated antigenic sites, ranking functionally critical residues far more effectively than traditional metrics. Correlations between SHAP values and standard evolutionary metrics were strong (0.63≤ρ≤0.9). These results demonstrate that explainable ML can reveal important substitutions, deliver tree free molecular dating and may transform passage metadata from a nuisance into an experimental probe.