LeadFinder machine learning predictions
LeadFinder, h2o.ai, R
default docking parameters
Ligands from ChemBl database and D3R 2017 cathepsin affinity ranking were docked with LeadFinder. Energy components of 10 best-score structures from independent runs were used to train an ensemble learning model in h2o.ai machine learning software to predict pIC50. Then the model was used to predict pIC50 for the new ligands (docked identically with LeadFinder) and the predicted pIC50s were used to rank the ligands.
No