Post date: Apr 19, 2015 9:55:15 PM
A draft version of spatpg (a program to estimate Ne and selection coefficients from spatio-temporal genetic data) is working. I will describe it in a future post, but here are the details for the first MCMC runs with a test data set (5 chains like this):
cd /local/scratch/
sleep 0
/home/A01963476/bin/spatpg -g /labs/evolution/projects/spatgen/qnemoSims/geno_sim0.txt -e /labs/evolution/projects/spatgen/qnemoSims/env_sim0.txt -n 100000 -b 20000 -t 80 -o out_sim0rep0.hdf5
scp out_sim0rep0.hdf5 /labs/evolution/projects/spatgen/mcmc/
And here is information from the mcmc.C source file on proposal distributions and priors:
// proposal distribution paramters
propNe = 50;
propP = 0.005;
propAlpha = 0.05;
propBeta = 0.05;
//prior parameters
ubNe = 2000;
lbNe = 20;
sdSlabAlpha = 0.5;
sdSlabBeta = 0.5;
lbSpikeAlpha = -0.01;
ubSpikeAlpha = 0.01;
lbSpikeBeta = -0.01;
ubSpikeBeta = 0.01;
spikeMixAlpha = 0.0;
spikeMixBeta = 0.0;