NEAT-POCKET: Pocket-Conditioned Autoregressive 3D Molecular Generation with a Neighborhood-Guided Set Transformer

Le résumé fourni par la source

trained_models.zip: Model weights for NEAT-POCKET, an autoregressive 3D molecular generation model, conditioned on protein pockets. We provide two versions: one trained on the CrossDocked2020 dataset, and the other one on the SPINDR dataset. generated_data.zip: Sampled molecules to reproduce our evaluation results. We additionally provide sampled molecules for the following baselines: Pocket2Mol, TargetDiff, DiffSBDD, DrugFlow, FLOWR.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.