Post date: Jan 07, 2016 5:10:45 PM
Here we only consider the cases of L = 10 or 100 with functional variants present (still h2 = 0.3 or 0.05). My main question is whether we can do substantially better with larger samples, such as 2500 individuals. This is big, but not crazy. The new files were simulated with sims2500.R and are labelled as data sets 2500_simsX where x ={1,2,3,4} and correspond to the same conditions as here. MCMC conditions also match those described previously. Here is an example.
perl wrap_qsub_rc_gemma.pl /labs/evolution/projects/selection_sims/geno2500_sims.txt /labs/evolution/projects/selection_sims/pheno2500_sims1.txt
cd /pscratch/A01963476/sims/gemmasel
sleep 5
gemma -g /labs/evolution/projects/selection_sims/geno2500_sims.txt -p /labs/evolution/projects/selection_sims/pheno2500_sims1.txt -bslmm 1 -n 50 -o out_sim1ph2500-50rep1 -rpace 100 -maf 0.0 -w 1000000 -s 2000000