GoldMine enables the interactive analysis and post-processing of GOLD docking results.
Scoring functions alone cannot be relied upon for identification of actives. GoldMine allows you to calculate a wide range of highly customisable descriptors that characterise how each docked ligand is interacting with the protein binding site. These quantify, amongst other things:
- simple properties such as the number of H-bonding ligand atoms, molecular weight of ligand, number of rotatable bonds
- the presence or absence of hydrogen-bonds between the protein and docked ligand
- the buried surface area of the ligand
- whether particular regions, or sub-pockets of the binding site are occupied by the ligand
Descriptors may then be used, e.g., to filter out unpromising ligands and to facilitate selection of suitable drug leads from virtual screens.
GoldMine allows easy visual inspection of GOLD docking results and interactive histograms and scatter plots can be generated for any numeric descriptors. The visualisation of interaction hot spots for active molecules can be used to identify regions of the binding site favoured by certain ligand atom types. The calculation and display of per-atom scores can be used to further analyse docking results.
GoldMine can be used to combine and analyse several docking runs, e.g. docking runs carried out against different protein models may be combined within a GoldMine database and analysed for selectivity and specificity. Docking poses scored using different scoring functions may also be combined and evaluated within a GoldMine database.
Enrichment metrics (e.g. AUC, ROC, BEDROC) can be generated interactively for virtual screens. Scoring function components and other descriptors may be combined using multiple regression to create discrimination models, target-specific scoring functions, and 3D QSAR models.