In an effort to take advantage of the increasing pervasiveness of PCs with multicore processors and to help researchers thwart the high cost of software-based accelerators, researchers at the University of Edinburgh Medical School reached out to Jon Hill for some serious help using R. "They analyze genomic data and were starting to hit what their hardware could run in serial using R," says Hill, an applications consultant at Edinburgh Parallel Computing Centre. "They were running out of memory and processing their data — just to do the initial data checking would take three or four days of programming time. … So they came to us and said, 'Look, we're running out of memory and computational power … can you give us a hand?'"
Hill and his colleagues recently unveiled their answer to this cry for help with the Simple Parallel R INTerface, or SPRINT, an open source solution that allows users take advantage of multicore computing to accelerate their R modules. While there is no lack of commercially available software package tools that parallelize R modules, affordability is a real concern, Hill says.
But even more than cost, a driving factor for the project was ease of use for bench biologists who aren't experts in computer science. "We're writing parallelized programming functions underneath using MPI and C, but an HPC programmer is doing this — and the biologist that wants to use R can just edit those scripts slightly and get full parallelization, so essentially we're [providing] the building block for biologists as a community," Hill says.
The fact that their BMC Bioinformatics paper introducing SPRINT was so highly accessed took Hill by surprise, but he thinks it may point toward the fact that when it comes to useable software solutions, biologists like to support open source tools. "I spoke to quite a few researchers around the university and they all seemed very keen to see this software blossom and get some use. But I really didn't expect it to be this highly accessed," he says. "But I think it's this idea of: don't change your existing scripts, let someone else who is an expert in HPC programming take care of the hard stuff for you."