I am the principal developer of this efficient and scalable open-source GEM software (Gene-Environment interaction analysis in Millions of samples). I am also involved in developing the statistical method. The source code can be found here. This software also supports cloud workflows. Our manuscript is also online.

The currently released GEM adopts a multi-threading application via boost thread management. GEM can also be converted to another multi-threading paradigm, i.e., OpenMP. Paralleling the most time-consuming hot spot when processing and analyzing the data sets is the main driver for choosing one multi-threading approach over the other. So, understanding your large-scale data sets is of priority. Let me know if you are interested in learning more. We can also work together to further improve GEM efficiencies in heterogeneous computing systems.