On August 8, 2022 Modulus Discovery, Inc., a computation-driven drug discovery company, reported the signing of a research collaboration with Fujitsu Limited and the unveiling of its proprietary computational technology platform, ModBind (Press release, Modulus Discovery, AUG 8, 2022, View Source [SID1234617733]). The collaboration will accelerate the large-scale deployment of Modulus discovery tools across world-leading supercomputers, including the supercomputer "Fugaku"(*1), and AI Bridging Cloud Infrastructure (ABCI, *2).
Schedule your 30 min Free 1stOncology Demo!
Discover why more than 1,500 members use 1stOncology™ to excel in:
Early/Late Stage Pipeline Development - Target Scouting - Clinical Biomarkers - Indication Selection & Expansion - BD&L Contacts - Conference Reports - Combinatorial Drug Settings - Companion Diagnostics - Drug Repositioning - First-in-class Analysis - Competitive Analysis - Deals & Licensing
Schedule Your 30 min Free Demo!
ModBind is a complementary addition to Modulus’ cutting-edge drug discovery platform that includes large-scale GPU-accelerated physics-based simulations, large-scale virtual screening algorithms, and specialized machine learning models. ModBind is a dynamic molecular simulation-based predictor of ligand efficacy, based on a theoretical approach that is fundamentally different from other simulation-based technologies on the market. This allows for up to 100-fold higher speed / throughput, the ability to predict efficacy for highly diverse / unrelated compound sets, higher reliability and repeatability of predictions, and accuracy comparable to the leading available technologies.
ModBind was originally developed by the Modulus team and is now being actively deployed to drive a number of the discovery programs in the company’s portfolio. Its accuracy has been demonstrated through internal validation studies that showed high levels of correlation with experimental measurements on publicly available diverse datasets for multiple relevant drug targets. The validation on these targets gave a correlation coefficient R2 = 0.67 and an average error = 0.51 log units. Because the technology is able to predict efficacy of ligand sets with highly diverse and unrelated chemical structures, this level of accuracy makes the technology suitable for use in higher accuracy medium-throughput virtual screening (VS) or post-VS filtering, as well as lead optimization of drug candidate molecules.
Modulus is working with Fujitsu to further optimize the performance, scalability, and deployment across multiple supercomputing environments for use in large scale virtual screening, compound library predictions, and in combination with generative methods to accelerate lead optimization. Fujitsu has already applied its HPC acceleration technologies to the MD simulation component of ModBind, implementing the method on a number of computing environments and has reproduced the original validation studies.