Performance comparison of Multifactor Dimensionality Reduction: JVM vs. Native code

Multifactor Dimensionality Reduction (MDR) was originally developed to detect gene-gene interactions in the domain of human genetics. MDR algorithm collapses high-dimensional genetic data, into a single dimension thus permitting interactions to be detected in relatively small sample sizes. To deal with the large volume of data, we demonstrate the experiment of MDR software which was written in Java programming language and compare with native method. We focus on execution time, memory usage and CPU usage and describe the results.