The ClusPro server and its PIPER docking engine are among the main products of our lab, and hence we are happy to see the increasing popularity of the server. According to the CAPRI docking experiment, ClusPro consistently has been among the best docking servers, and we expected that it will be heavily used, but the actual usage exceeded our wildest dreams. In 2019 the server performed 98,300 docking calculations, and in the first 8 months of 2020 we already recorded over 119,000 docking jobs! Even in August, which used to be a slow month, we performed over 20,000 dockings. It is possible that due to the pandemic researchers could not take any vacation and chose to work instead, but we have never seen such numbers before.
Of course we love to see people using our software, but having so many users also means increased responsibility and causes us two major concerns. First, we have to assure smooth operation of the server, and we are in the process of engaging outside help for developing an API (Application Programming Interface) for improved scheduling of the submitted jobs. Depending on the availability of funds we also plan to add new buy-in computer resources. Second, in spite of being among the best docking servers, ClusPro has many limitations as described in our just published paper (1), and we are hard at work to increase the robustness of the method to solve a higher fraction of docking problems and improve the accuracy of the models. In particular, we already developed a machine learning based method for improved ranking of the clusters of predicted structures, and expect to add it to the server in the near future.
I. T. Desta, K. A. Porter, B. Xia, D. Kozakov, S. Vajda, Performance and Its Limits in Rigid Body Protein-Protein Docking. Structure 10.1016/j.str.2020.06.006 (2020).