PepEVOLVE is a position-aware, dynamic framework for peptide lead optimization that learns both where to edit a peptide and how to optimize it for multi-parameter improvement. It addresses a key limitation of prior generative approaches like PepINVENT, which require chemists to pre-specify mutable positions that are not always known a priori and rely on static pretraining and optimization algorithms.
