Skip to main content
Fig. 9 | Journal of Cheminformatics

Fig. 9

From: Applying atomistic neural networks to bias conformer ensembles towards bioactive-like conformations

Fig. 9

Inspection of a ComENet failure case of early retrieval of bioactive-like conformations for the MT3 ligand in the 3efj complex. A Actual ARMSDbio compared to ComENet prediction. B ARMSDbio compared to the Sage energy. C ARMSDbio compared to the TFD to the MCS of the closest molecule in training set. D Comparison between the ComENet predictions of the generated conformers of the test molecule and the predictions of generated conformers of the closest molecule based on matching conformers using torsion angles of their MCS. E Overlay between the bioactive pose of the MT3 ligand in the 3efj complex (green) and closest ligand AM7 binding the same protein pocket in the 2rfn complex (cyan)

Back to article page