The abstract is read at the publisher; the summary is JClub's.
SMSTPred predicts Class I SAM-dependent methyltransferase substrate type with a cluster-disjoint Matthews correlation coefficient of 0.94 across ~116,000 MTases.
- Why it matters: Most SAM-MTases have unknown substrate types in sequence databases, limiting functional annotation and obscuring how methylation specialization evolves across life.
- What they did: The multi-label classifier uses protein language model embeddings and is benchmarked under homology-disjoint cross-validation against MMseqs2, Foldseek, and profile-HMM baselines.
- The result: Predicted substrate specialization is linked to taxonomy, with small-molecule methyltransferases enriched in plants and Actinomycetota, and it assigns substrate types to thousands of uncharacterized enzymes.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.