FINDSITEcomb2.0: A Drug Discovery Tool is an upgrade of the FINDSITEcomb algorithm that was compared to several commercially and freely available docking programs against the DUD set – A Directory of Useful Decoys. We demonstrated that FINDSITEcomb has virtual screening accuracy better than the best docking method under the challenging condition that no templates > 30% sequence ID to the target are present in the ligand binding databases. FINDSITEcomb2.0 is shown to have significant improvment over FINDSITEcomb. FINDSITEcomb2.0 also outperforms state-of-the-art methods that employ machine learning such as a deep convolutional neural network, CNN. With 80% sequence identity cutoff of target to templates for DUD-E set and modeled target structures, the average AUC and ROC 1%, ROC-EF1 enrichment factor of FINDSITEcomb2.0 are 0.879 & 53.72, compared to 0.868 & 29.65 of the deep CNN scoring method, respectively, which employed crystal structures. If FINDSITEcomb2.0 uses crystal structures, then the AUC and ROC-EF1 are 0.898 and 58.22, respectively. Thus, FINDSITEcomb2.0 represents a significant improvement in the state of the art.
Notice: This server is free for non-commercial users.
Commercial users – to use this server, please send an email to Dr. Jeffrey Skolnick: skolnick@gatech.edu.
If you find this service useful, please cite the following paper: Zhou, H, Cao H, Skolnick J. 2018. FINDSITEcomb2.0: A New Approach for Virtual Ligand Screening of Proteins and Virtual Target Screening of Biomolecules. Journal of Chemical Information and Modeling. 58(11): 2343-2354 doi: 10.1021/acs.jcim.8b00309. PDF
For virtual ligand screening
Output format
Download benchmarking results
This site is maintained by Dr. H. Zhou: hzhou3@gatech.edu