Hardware acceleration of Generalized Born for simulation based drug design and molecular modeling of proteins

• Research unit: INNOSUISSE
• Project number: 8808.1;7 PFLS-LS

Short description (English)

Hardware acceleration of Generalized Born for simulation based drug design and molecular modeling of proteins

Short description (French)

Hardware acceleration of Generalized Born for simulation based drug design and molecular modeling of proteins

Abstract (French)

Bio-simulations of proteins bring key insights regarding their conformational mode and their interaction with cellular membranes with a rational approach. Alternatively, such simulations are used to study the binding modes of synthetic compounds on protein active sites. These simulations are part of the molecular modeling field. The molecular modeling field opens new avenues for identifying new target based on gene sequences, identifying and optimizing leads. The full potential of molecular modeling is not achieved because of the limited computing power of currently available computers, the complexity and costs related to deploying large computing centers. Indeed simulations of several months are needed to study the behavior of small proteins in a 100 nanosecond time scale whereas several microseconds are needed to reach strong predictive biological significance. The UNIL-SIB Molecular Modeling Group and XLBiosim SA join forces to develop and validate a new hardware based acceleration system for drug design and protein modeling. This product is based on Generalized Born model which is widely used in the molecular modeling community, for applied as well as fundamental research. It includes a custom hardware chip that is integrated with the widely used software package CHARMm. This product will deliver 10x speedups for molecular simulation based on Generalized Born when compared to commercial processors of the same silicon technology generation.