LassoESM is a language model specifically tailored for lasso peptides, designed to improve the prediction of their properties.
It utilizes a domain adaptation approach to further pre-train ESM-2 (650 million parameters) model on lasso peptide datasets using the masked language modeling technique.
