1146-1-o2jtn-PosePredictionProtocol.txt

Name

Multiple initial conformations with manual and ligand similarity based selection from the top-50 poses.

Software

AutoDock Vina with in-house modifications, RDKit 2017, SciPy, LIGSIF, PyMOL 1.8.4, unreleased version of Convex-PL scoring function

System Preparation Parameters

1000 conformations for each molecule in RDKit, sequence identity for proteins >= 99

System Preparation Method

BACE 1 - 19 conformations were generated in RDKit and clusterized with hierarchical clustering. Each 100 clusters were selected for docking. Receptors downloaded from RCSB with help of their search by sequence feature.

Pose Prediction Parameters

Vina exhaustiveness 10, number of poses 350-450, Convex-PL rescoring with typization system 14, regression-based model with grid-based solvent volume contribution, regression-based model with SASA descriptors.

Pose Prediction Method

We ran docking experiments with 100 conformations of each ligand except BACE_20, did the rescoring with several versions of Convex-PL, and, since it turned out that the 100 conformations were often not enough to model properly the flexibility of the ring, had to reject some poses from the top of predictions based on the ligand similarity criterion (visually and numerically by computing the rmsd between the maximum common substructures).

Answer 1

Yes

Answer 2

Yes