Study on the Structure and Properties of Protein-Ligand Complexes

Swiftly screening the best ligands and docking poses often leads to long processing times or loss of molecular information. Thus, inspired by the search for the top-k semantically similar sentences in the field of Natural Language Processing (NLP), I contributed to the development of the MapLE method, a general framework for Matching molecular analogues promptly with Low computational resources by multi-metrics Evaluation. This approach integrates various similarity metrics and introduces a progressive prompt evaluation technique to expedite the screening process. Experimental validation on a public biomolecular dataset confirms the effectiveness and efficiency of this strategy.