Post date: Aug 28, 2014 10:5:20 PM
The LMM analyses finished. There is a single SNP significantly associated with survival in C and none for A at the genome wide significance level (p/#loci). There are many others with relatively low p-values. For A, the REMLE and MLE estimates of lamda are 0.001 and the pve is near zero with a giant s.e. (about 0.5). Results for C are similar, except that the MLE for lambda is quite a bit higher (but not the REMLE). I am not sure what to think about this, the pve and lambda estimates in particular. I wonder whether the n <<< p problem is simply too great here. I might re-run some of the analyses with only loci with maf > 5% to reduce p.