Post date: Dec 14, 2015 3:25:50 AM
My initial run of popanc with the mouse data sets was incorrect as I allowed more SNPs in a window than reference individuals. This violates LDA. I am re-running it with +- 4 SNPs from the focal SNP (9 total), which is fewer SNPs than the 5 (P0) + 5 (P1) + 5 (F1) = 15 reference individuals. I also fixed a bug that very rarely resulted in an index out of bounds if a maximum eigenvalue could not be found. And I added a small automatic jitter to genotypes to avoid these being identical across SNPs and individuals (normal deviated with s.d. 1e-9). I might make this an option, but its such a small value that there is no harm (that I can think of) anyway. Here is an example of the re-runs:
cd /local/scratch/
sleep 10
popanc -o estpanc_n4musTUChain2.hdf5 -m 30000 -b 10000 -t 5 -n 4 -d 10 -s 1 -u 20 -l 0.05 -a 0.1 -z 1 /labs/evolution/projects/popanc_sims/mouse/infiles/genoP0.txt /labs/evolution/projects/popanc_sims/mouse/infiles/genoP1.txt /labs/evolution/projects/popanc_sims/mouse/infiles/genoTU.txt
scp estpanc_n4musTUChain2.hdf5 /labs/evolution/projects/popanc_sims/mcmc/mus/