Post date: Nov 22, 2013 4:49:8 PM
I generated five permutations of the survival phenotype for SLA on Ac. This is the experiment where we have very high pve. I wanted to see whether this is simply caused by having lots of data and a skewed, binary phenotype (most individuals lived). I used gemma to estimate the pve for these five data sets. The pve (point estimate) ranged from 0.78-0.99. Thus, the pve range for the permuted data sets includes the pve for the observed data. This means we should be very cautious in our interpretation of this pve. In other words, it might be rather meaningless.