Report from NESCent working group: SimBank

Post date: Nov 15, 2014 6:37:27 PM

Posted by Dr. L:

Last week I attended an NESCent working group called SimBank, organized by Mike Whitlock, Victoria Sork, and Matt Hahn. Our goal was to explore the possibility of creating a standardized set of simulations that can be used to test population-genetic methods, and to house them in a central repository where other simulations could be uploaded. After a day or so of serious conversation, we settled in on the idea of creating a set of 10-12 "benchmark" simulated datasets that could be used to challenge methods---rather than the simple simulations that are commonly used. Inspired by model organisms such as humans, Drosophila, and Arabidopsis, and non-model organisms such as conifers, urchins, and whitefish, we drafted a series of demographic histories that should challenge most methods. The idea is to simulate the same genetic architecture on each demography.

Attaining our goals will be challenging: we have to find a centralized repository; someone has to conduct the simulations and make sure they are reproducible; and we have to create momentum among our colleagues to use these datasets for testing their programs. Keep an eye out for our paper that will summarize these goals and outline the plan of action for SimBank.