Work

Extensive SARS-CoV-2 spike sequence diversification converges on a bounded, low-dimensional physicochemical space

Year: 2026
Type: preprint
Abstract: During the COVID-19 pandemic, SARS-CoV-2 accumulated extensive amino acid variation in its spike protein. However, the extent to which sequence diversification translates into physicochemical diversity remains unclear. Here, we analyzed 271,769 unique spike protein sequences from 12 geographical regions using three physicochemical descriptors: net charge at pH 7.0, hydropathy, and aliphatic index. To characterize temporal changes in physicochemical space, we compared the squared Mahalanobis distances of observed sequences with those of Monte Carlo-simulated sequences containing randomized amino acid substitutions, constrained to the number of empirically variable positions. We applied trimmed clustering to characterize sequence distributions in physicochemical space and used Shannon entropy to assess spatiotemporal relationships between genotypic and physicochemical diversity across geographic regions. Empirical physicochemical displacement increased early in the observation period and subsequently stabilized, despite continued increases in variable amino acid positions and accumulated substitutions. In contrast, simulated trajectories exhibited greater and sustained physicochemical divergence. Trimmed clustering identified six distinct physicochemical groups repeatedly occupied by genetically divergent lineages differing in mutation burden, amino acid composition, and emergence time. Shannon entropy analyses revealed broadly similar nonlinear temporal patterns across geographic regions, with differences between genotypic and physicochemical entropy tending to be negative. These findings indicate that extensive spike protein sequence diversification occurred within a comparatively restricted physicochemical space, consistent with a many-to-one relationship between amino acid sequences and physicochemical states. This framework provides a quantitative approach for investigating physicochemical constraints on viral protein evolution and may inform future machine learning approaches to epitope prediction and vaccine antigen design.
Language: English
Field: Medicine
Open Access status: green
OpenAlex ID: W7233776393
Cites: 55
Cited by:
Related to: 10