cuTauLeaping: A GPU-Powered Tau-Leaping Stochastic Simulator for Massive Parallel Analyses of Biological Systems

Abstract

Tau-leaping is a stochastic simulation algorithm that efficiently reconstructs the temporal evolution of biological systems, modeled according to the stochastic formulation of chemical kinetics. The analysis of dynamical properties of these systems in physiological and perturbed conditions usually requires the execution of a large number of simulations, leading to high computational costs. Since each simulation can be executed independently from the others, a massive parallelization of tau-leaping can bring to relevant reductions of the overall running time. The emerging field of General Purpose Graphic Processing Units (GPGPU) provides power-efficient high-performance computing at a relatively low cost. In this work we introduce cuTauLeaping, a stochastic simulator of biological systems that makes use of GPGPU computing to execute multiple parallel tau-leaping simulations, by fully exploiting the Nvidia's Fermi GPU architecture. We show how a considerable computational speedup is achieved on GPU by partitioning the execution of tau-leaping into multiple separated phases, and we describe how to avoid some implementation pitfalls related to the scarcity of memory resources on the GPU streaming multiprocessors. Our results show that cuTauLeaping largely outperforms the CPU-based tau-leaping implementation when the number of parallel simulations increases, with a break-even directly depending on the size of the biological system and on the complexity of its emergent dynamics. In particular, cuTauLeaping is exploited to investigate the probability distribution of bistable states in the Schlögl model, and to carry out a bidimensional parameter sweep analysis to study the oscillatory regimes in the Ras/cAMP/PKA pathway in S. cerevisiae.

LASSIE: simulating large-scale models of biochemical systems on GPUs

“…The models were generated considering the methodology used in [30, 50], which was modified in order to randomly sample the initial concentration of each species with a uniform distribution in the range [0,1), and the kinetic constant of each reaction with a logarithmic distribution in the range [10 −8 ,1).…”

Towards Human Cell Simulation

“…Parallel computing paradigm may be used on multi-core CPUs, many-core processing units (such as, GPUs [77]), re-configurable hardware platforms (such as, FPGAs), or over distributed infrastructure (such as, cluster, Grid, or Cloud). While multi-core CPUs are suitable for general-purpose tasks, many-core processors (such as the Intel Xeon Phi [24] or GPU [85]) comprise a larger number of lower frequency cores and perform well on scalable applications (such as, DNA sequence analysis [71], biochemical simulation [53,76,81,123] or deep learning [129]).…”

“…3.1 for some examples). In this context, GPUs [77] were already successfully employed to achieve a considerable reduction in the computational times required by the simulation of both deterministic [53,76,123] and stochastic models [81,150]. Besides accelerating single simulations of such models, these methods prove to be particularly useful when there is a need of running multiple independent simulations of the same model.…”

Gillespie’s Stochastic Simulation Algorithm on MIC coprocessors

“…Algorithm parallelization is usually realized by means of multi-threading [23], distributed computing on clusters [10], custom circuitry produced with Field Programmable Gate Array (FPGA) [14] or general-purpose Graphics Processing Units (GPU) computing [17,18,20]. These parallel technologies generally require a custom implementation of the algorithm, since most of the time CPU code cannot be directly ported on the parallel architecture; in addition, distributed architectures need the definition of an appropriate scheduler to manage the parallel execution of processes.…”

“…The second test case is a family of synthetic stochastic models of increasing size (SynSM, in short), which are randomly generated according to the methodology proposed in [20]. Namely, SynSM are characterized by a number of species N and of reactions M ranging from 20 × 20 to 240 × 240; the values of the stochastic constants are randomly sampled with uniform distribution in (0, 1).…”