FRAGSITE is a fragment-based machine learning algorithm that scores protein-ligand binding for ligand virtual screening. FRAGSITE exploits our observation that ligand fragments, e.g. rings, tend to interact with stereochemically conserved protein subpockets that also occur in evolutionarily unrelated proteins. FRAGSITE was benchmarked on the 102 protein DUD-E set, where any template protein whose sequence identify >30% to the target was excluded. Within the top 100 ranked molecules, FRAGSITE improves VLS precision and recall by 14.3% and 18.5% respectively relative to FINDSITEcomb2.0. Moreover, the mean top 1% enrichment factor increased from 25.22 to 30.20 (an average increase of 19.8%). Experimental testing of FRAGSITE’s predictions show that FRAGSITE discovers more hits and covers a more diverse region of chemical space than FINDSITEcomb2.0. For the two proteins tested, DHFR, a well-studied protein that catalyzes the conversion of dihydrofolate to tetrahydrofolate, and ACVR1, a kinase implicated in DIPG, a childhood brain cancer, FRAGSITE identified novel small molecule nanomolar binders. Interestingly, one novel binders of DHFR is a kinase inhibitor which is predicted to bind in a novel subpocket. For ACVR1, FRAGSITE identified novel molecules that have diverse scaffolds and estimated nanomolar to micromolar affinities. Thus, FRAGSITE shows a significant improvement over prior state-of-the-art ligand virtual screening approaches.
NOTE:
- This web service is freely available to all academic users and not-for-profit institutions.
- Commercial users wishing an evaluation copy should contact skolnick@gatech.edu.
If you find this service useful, please cite the following paper: Zhou, H, Cao H, Skolnick J. 2021. FRAGSITE: A Fragment Based Approach for Virtual Ligand Screening. J Chem Inf Model. 61(4): 2074-2089. doi: 10.1021/acs.jcim.0c01160. PDF