1147-1-knouw-PosePredictionProtocol-KF.txt

Name

Deep learning assisted similarity docking

Software

Efindsite 1.2, Openbabel 2.4.1, Discover studio visualizer 4.5, Maestro 10.2, Shafts, Guassian 09, Amber 16, Homemade deep learning

System Preparation Parameters

None

System Preparation Method

The template dataset was generated by efindsite. Receptors were prepared using Discover studio visualizer.

Pose Prediction Parameters

None

Pose Prediction Method

1) Efindsite was used to find out all similar protein-ligand complexes to the target from pdb database, and based on that, a template set was built.
2) For each target ligand, using openbabel, the similarity scores to all ligands in this template set were calculated, and then ranked, the protein-ligand complexes with high similarities were picked up from the template set.
3) Based on templates with high similarities, the pose candidates were built through some manual adjustment implemented in Discover studio visualizer and Maestro.
4) Some minor optimizations to the pose conformations were performed by Guassian as well as Amber.
5) Pose candidates were ranked by homemade deep learning algorithms.

Answer 1

Yes

Answer 2

Yes

1479-2-5u76u-LigandScoringProtocol-MLCL.txt

Name

Deep-Learning-Package-MLCL

Software

AG/DG/TDL-BP/Schrodinger

Parameters

Use AG, DG and/or TDL-BP with default parameters
Use PDBBind as the training data for these models

Method

The features generated by Algebraic Graph, Differential geometry, and Algebraic topology scores are utilized in in-house design neural network models to predict the binding free energy

Answer 1

No

Answer 2

Yes