Target-specific Grid-Rescoring, trainingThis tool generates a model for Target-specific Grid-Rescoring (TaGRes).
As input we need:
* a file containing a protonated protein in pdb-format
* a file containing a reference ligand. This reference ligand should be located in the binding pocket. Supported formats are mol2, sdf or drf (DockResultFile, xml-based).
* a file containing a training data set, i.e. compounds whose binding-free-energy to the specified target is known and annotated in this file. Those compounds should have been docked into the specified protein.
A scoring function is applied and an interaction-grid is thereby generated for each input compound. Together with the known binding-free-energy, those grids are used to automatically search for the best linear or non-linear regression model that can approximate the binding-free-energy. After this model has been generated, you can pass it to the tool TaGRes and rescore (different) compounds with it.
The output of TaGRes-train is a file that contains the generated regression model. However, if no model with suitable prediction quality was found, an error will be shown and no model-file will be written.