Accelerating gene regulatory networks inference through GPU/CUDA programming

摘要: Gene regulatory networks (GRN) inference is an important bioinformatics problem in which the gene interactions need to be deduced from expression data, such as microarray data. Feature selection methods can applied this problem. A feature technique composed by two parts: a search algorithm and criterion function. Among algorithms already proposed, there exhaustive where best subset returned, although its computational complexity unfeasible almost all situations. The objective of work development low cost parallel solution based on GPU architectures for with viable cost-benefit. CUDA™ general purpose architecture new programming model allowing that NVIDIA® GPUs solve complex problems efficient way. We developed GRN GPU/CUDA encouraging speedups (60x) were achieved when assuming each target has predictors. idea behind proposed method considering three or more predictors well.