Parallel information theory based construction of gene regulatory networks

We present a parallel method for construction of gene regulatory networks from large-scale gene expression data. Our method integrates mutual information, data processing inequality and statistical testing to detect significant dependencies between genes, and efficiently exploits parallelism inherent in such computations. We present a novel method to carry out permutation testing for assessing statistical significance while reducing its computational complexity by a factor of Θ(n ²), where n is the number of genes. Using both synthetic and known regulatory networks, we show that our method produces networks of quality similar to ARACNE, a widely used mutual information based method. We present a parallelization of the algorithm that, for the first time, allows construction of whole genome networks from thousands of microarray experiments using rigorous mutual information based methodology. We report the construction of a 15,147 gene network of the plant Arabidopsis thaliana from 2,996 microarray experiments on a 2,048-CPU Blue Gene/L in 45 minutes, thus addressing a grand challenge problem in the NSF Arabidopsis 2010 initiative.