Parallel mutual information estimation for inferring gene regulatory networks on GPUs

Abstract

BackgroundMutual information is a measure of similarity between two variables. It has been widely used in various application domains including computational biology, machine learning, statistics, image processing, and financial computing. Previously used simple histogram based mutual information estimators lack the precision in quality compared to kernel based methods. The recently introduced B-spline function based mutual information estimation method is competitive to the kernel based methods in terms of quality but at a lower computational complexity.ResultsWe present a new approach to accelerate the B-spline function based mutual information estimation algorithm with commodity graphics hardware. To derive an efficient mapping onto this type of architecture, we have used the Compute Unified Device Architecture (CUDA) programming model to design and implement a new parallel algorithm. Our implementation, called CUDA-MI, can achieve speedups of up to 82 using double precision on a single GPU compared to a multi-threaded implementation on a quad-core CPU for large microarray datasets. We have used the results obtained by CUDA-MI to infer gene regulatory networks (GRNs) from microarray data. The comparisons to existing methods including ARACNE and TINGe show that CUDA-MI produces GRNs of higher quality in less time.ConclusionsCUDA-MI is publicly available open-source software, written in CUDA and C++ programming languages. It obtains significant speedup over sequential multi-threaded implementation by fully exploiting the compute capability of commonly used CUDA-enabled low-cost GPUs.

Parallel Mutual Information Based Construction of Whole-Genome Networks on the Intel (R) Xeon Phi (TM) Coprocessor

“…Out of these, CUDA-MI [9] is the most recent and the fastest reported performance. Using Nvidia Tesla C2050 GPU, the authors report computing pairwise MI values for 10000 genes and 4000 observations in 838.74 seconds.…”

“…There has been recent work on the parallelization of MI on GPUs [9], [12], [13]. Out of these, CUDA-MI [9] is the most recent and the fastest reported performance.…”

“…Apart from the specific problem being solved, this work is useful in two additional ways. Mutual information is a widely used technique with its roots in information theory, and there has been recent work on its parallelization for GPUs [12], [13], [9]. Our work is the first implementation of a B-Spline based mutual information kernel on the Intel Xeon Phi coprocessor, and is useful for porting other MI based applications to this architecture.…”

“…In works that assess different methods for network construction (see, for example [1], [2]), a compelling observation is that simpler methods such as those that only decipher pairwise linear interactions are computationally expedient, whereas higher quality methods including mutual information [3] and Bayesian approaches [4] face scaling limitations. This has spurred research into the development of parallel methods [5], [6], [7], [8] and GPU implementations [9].…”

Information-theoretic analysis of molecular (co)evolution using graphics processing units

“…While general MI computations have been implemented on GPUs before [22,23], the above mentioned normalization has not been done for coevolutionary biosequence analysis: Shi et. al.…”