Genetic variation shapes human health, yet the sheer number of mutations across proteins—often thousands per gene—remains poorly understood. Only a small fraction has been experimentally studied, and available knowledge is scattered in many places. This fragmentation creates a major barrier to linking mutations with disease mechanisms or drug response, slowing progress in biomedical research. CATVariant is a powerful, interactive webserver that unifies sequence, structure, function, clinical evidence, and drug response predictions to streamline the study of protein mutations. Provided the protein name, CATVariant identifies the wild-type sequence and unifies structural features, variants, population and demographic data, clinical evidence, experimental results, predictive analyses, and references into a single interactive dashboard. Its built-in scoring system synthesizes these evidence streams to prioritize variants, and in validation studies it successfully flagged numerous mutations in key drug targets later confirmed as pathogenic in clinical reports. Users can explore mutations in a sequence viewer or 3D models, filter by ethnicity or sex, and focus on variants most likely to have functional impact. Furthermore, CATVariant incorporates a deep learning workflow integrating Boltz-2 model to predict how mutations influence drug response within minutes, generating protein-drug complex structures, estimating binding affinities, and revealing how the mutations alter molecular interactions at the binding site. Predictions show strong agreement with experimental data ( r = 0.71, R 2 = 0.51 across 20 drugs on hERG mutants; r = 0.65, R 2 = 0.43 across 18 drugs on Na V 1.5 mutants), surpassing traditional methods in speed and accessibility. By unifying variant discovery, functional characterization, and pharmacological prediction, CATVariant provides a universal, accessible resource for any protein of interest. It empowers researchers to identify pathogenic mutations, understand demographic distribution, and forecast drug responses where experiments are lacking—accelerating discovery, improving drug safety, and advancing precision medicine across the proteome.
Ngo et al. (Sun,) studied this question.