High Performance Computing for Detecting Complex Diseases using Deep Learning

The study of the Genome-wide association study (GWAS) and the complex diseases is of high importance nowadays. The epistasis describes the analysis of the single nucleotide polymorphisms (SNPs) interactions and their effects on the complex diseases. However, enormous number of SNPs interactions should be tested against the disease that is highly computational expensive. In this paper, High Performance Computing (HPC) is being applied on a supercomputer to reduce the processing time. Parallel Deep Learning (PDL) is applied and tested using different datasets. Simulated datasets of 12 different models and the real WTCCC Rheumatoid arthritis (RA) dataset are being tested. Results show the high accuracy, specificity and true positive rate values. Moreover, they show low values of the false discovery rate and the robustness of power through the different simulated models. When tested on the real RA dataset, our model shows the ability to detect the 2-way interaction SNPs with their promising related genes with high accuracy due to the parallel deep learning architecture.