Fig. 3
From: End-to-end protein–ligand complex structure generation with diffusion-based generative models
Modeling accuracies on the PocketMiner dataset. A Median modeling accuracies for each complex in the PocketMiner dataset, where the results by DPL, DPL+S (apo), and DPL+S (holo) are compared. B Fraction of complexes where a structure with L-rms below the threshold, varying from 0 Å to 5 Å, was sampled. The results obtained with GNINA are also shown for the apo and holo structures. C Distribution of L-rms for each complex in the PocketMiner dataset. The * on the PDB ID indicates that no related sample (\(\text {TM-score} > 0.5\)) was observed during the training