GPUs are massively multi-core accelerators designed for computer graphics that can be leveraged to distribute the calculations in scientific computing.
I worked on the acceleration of biochemical simulators, leveraging GPU’s cores to accelerate multiple simulations of a same model (“coarse-grained” acceleration) or to accelerate the simulation of a single massive model (“fine-grained” acceleration).
Fine + coarse-grained simulators
FiCoS
FiCoS is a GPU-powered deterministic biochemical simulator for models based on mass-action kinetics, based on a custom adaptive explicit/implicit integration algorithm. FiCoS is suitable for the simulation of multiple massive models.
More information about FiCoS is available on GITHUB.
Coarse-grained simulators
cupSODA
cupSODA is a GPU-powered deterministic biochemical simulator for models based on mass-action kinetics, based on the LSODA integration algorithm.
More information about cupSODA is available on GITHUB.
PySB/cupSODA
PySB is a framework for building mathematical models of biochemical systems as Python programs. PySB is a powerful tool for rule-based modeling and it is now integrated with the GPU-powered deterministic biochemical simulator cupSODA.
Read the paper about PySB/cupSODA on Bioinformatics
cuTauLeaping
cuTauLeaping is a GPU-powered stochastic biochemical simulator based on Cao’s version of the tau leaping algorithm.
More information about cuTauLeaping is available on GITHUB.
Fine-grained simulators
LASSIE
LASSIE is a GPU-powered deterministic biochemical simulator for large-scale models based on mass-action kinetics.
More information about LASSIE is available on GITHUB.