Skip to main content
. 2017 Apr 13;33(16):2513–2522. doi: 10.1093/bioinformatics/btx215

Fig. 1.

Fig. 1.

Computation time in minutes (y-axis) as a function of the number of SNPs d (c = 12, n = 600), the number of phenotypes c (d = 500, n = 600), and the number of subjects n (c = 12, d = 500). In each case, the computation time reported is based on 10 000 MCMC iterations (5000 iterations was a sufficient burn-in in all cases considered) with each run employing 49 cores (each 2.66-GHz Xeon x5650) on a computing cluster with 20 GB of RAM requested for each job. Each core is used to run the MCMC algorithm with a unique setting for the tuning parameters and a total of 49 settings are considered. Increasing or decreasing the number of settings, and hence the number of cores used, has no impact on the reported computation times