Post date: Dec 04, 2013 5:3:39 PM
I set up gemma analyses to predict survival using the probit model for 10 data sets with survival data dropped for 20% of individuals. Each analysis has a 1 million iteration burn-in and 5 million steps. These are running on the dorc cluster in the long queue. Here is an example of the commands,
gemma -g geno_glaTrt -p pheno_trainSurv.txt -bslmm 3 -n 10 -o survTrain_glaTrtSet10Ch1 -rpace 40 -w 1000000 -s 5000000
gemma -g geno_glaTrt -p pheno_trainSurv.txt -n 10 -gk 1 -o survTrain_glaTrtSet10Ch1
gemma -g geno_glaTrt -p pheno_trainSurv.txt -n 10 -epm ./output/survTrain_glaTrtSet10Ch1.param.txt -emu ./output/survTrain_glaTrtSet10Ch1.log.txt -ebv ./output/survTrain_glaTrtSet10Ch1.bv.txt -k ./output/survTrain_glaTrtSet10Ch1.cXX.txt -predict 2 -o survTrain_glaTrtSet10Ch1k