Lead Finder performs virtual screening of libraries of chemical compounds to find most potent binders for a given target protein. During virtual screening each ligand from a library is docked to a target protein, and ligands are rank-ordered according to their binding potencies.
To distinguish active compounds from the inactive ones more effectively, a special type of scoring function (called VS-score) has been implemented in Lead Finder. This scoring function employs the same energy contributions as other scoring functions of the Lead Finder (see Science
section for details), however the energy scaling coefficients have been specifically adjusted to attain better enrichment factors in virtual screening experiments.
To accelerate processing of large libraries in virtual screening experiments, Lead Finder employs a special modification of the docking algorithm called screening mode. Structure processing is on average 2-4 times faster in the screening mode than in the default docking mode. The fast screening mode does not lead to a significant loss of accuracy of free energy of binding prediction and docking success rate (see the Docking Success Rate
section for details).
Efficiency of Lead Finder in virtual screening has been benchmarked on a set of 34 therapeutically relevant protein targets for which crystallographic structures and sets of potent inhibitors are available from the public domain. As can be seen from the Virtual Screening Performance
section Lead Finder has been able to achieve impressive enrichments for almost all protein targets in the test set.