Post date: Feb 27, 2014 5:21:54 PM
I set up 10 replicate data sets for adult weight for each population x plant combination that dropped 20% of the data for cross-validation (this is only for survivors). The relevant R script is projects/lycaeides_hostplant/melGemma/scripts/makeWgtTestSets.R and the new pheontype files are projects/lycaeides_hostplant/melGemma/phen_trainWgt*.txt.
I using gemma to predict phenotypes (weight) for the individuals with dropped phenotypes. There are 80 runs total, 4 experiments, 10 data sets, and 2 MCMC runs each. These are going on the dorc cluster. Here is an example command:
cd /home/A01963476/projects/lycaeides_hostplant/melGemma/
sleep 5
gemma -g geno_slaTrtMs.txt -p pheno_trainWgtSlaMs.txt -bslmm 1 -n 10 -o wgtTrain_slaTrtMsSet10Ch1 -rpace 40 -w 1000000 -s 5000000
gemma -g geno_slaTrtMs.txt -p pheno_trainWgtSlaMs.txt -n 10 -gk 1 -o wgtTrain_slaTrtMsSet10Ch1
gemma -g geno_slaTrtMs.txt -p pheno_trainWgtSlaMs.txt -n 10 -epm ./output/wgtTrain_slaTrtMsSet10Ch1.param.txt -emu ./output/wgtTrain_slaTrtMsSet10Ch1.log.txt -ebv ./output/wgtTrain_slaTrtMsSet10Ch1.bv.txt -k ./output/wgtTrain_slaTrtMsSet10Ch1.cXX.txt -predict 1 -o wgtTrain_slaTrtMsSet10Ch1k