Affymetrix gene expression correlation with survival

Following the analysis of survival and prediction by other clinical traits here. I used the same survival object to investigate if there are any genes that are predictive of survival. The processed data was obtained from Bin (my understanding that Justin normalized it by removing batch (exact procedure of normalization needs to be specified)).

Permutation test:

P value	FDR, %	Number of loci
0.05	56.88	1048
0.01	45.23	264
0.005	42.99	144
0.001	32.62	48

So this doesn't look very convincing either. Need to do calculation for the progression free interval