MMseqs2

anchor

Findings

• IC-269Influence functions on ESM-2 650M reveal a power law tail in training data influence on sequence likelihood, with influence diminishing as Hamming distance from the wild type increases [supporting]
• IC-270Unsupervised finetuning (evo-tuning) on homologous sequences improves ESM-2 650M zero-shot fitness prediction for low-likelihood wild types but harms high-likelihood ones, with optimal threshold at log-likelihood ε = −1.4 [supporting]