SeqGSEA: High p-value, low FDR

[heah]

Hello!

I'm running SeqGSEA with 1000 permutations and the enriched gene sets have high p-values (lowest is 0.105) and low FDRs. Does this make sense? Shouldn't FDR be always higher than the corresponding p-value? Also, many of the p-values are the same.

I get these warnings when running the analysis:
4: In .local(object, ...) :
in estimateDispersions: sharingMode=='gene-est-only' will cause inflated numbers of false positives unless you have many replicates.
5: In parametricDispersionFit(means, disps) :
Dispersion fit did not converge.

Here's a sample output:

GSName GSSize ES ES.pos pval FDR FWER
GSE10325 168 1.22808662 5130 0.105 0 0.499
GSE12366 179 1.309364504 3994 0.105 0 0.391
GSE20366 168 1.245278455 4337 0.105 0 0.391
GSE22886 141 1.23398734 4313 0.105 0 0.499
GSE22886 176 1.244872228 3485 0.105 0 0.391
GSE24634 161 1.231951927 1188 0.105 0 0.499
GSE27786 186 1.231516643 2291 0.105 0 0.499
GSE39820 175 1.236275596 2472 0.105 0 0.499
GSE39820 174 1.250791386 3043 0.105 0 0.391
GSE17721 173 1.190668483 4093 0.105 0.0125 0.711
GSE3982 161 1.191279385 6536 0.105 0.012658 0.711
GSE3982 164 1.193291389 2810 0.105 0.012987 0.711
GSE7460 151 1.193674617 5536 0.105 0.013158 0.604
GSE360 154 1.195713581 4812 0.105 0.013514 0.604
GSE8515 168 1.195932546 2346 0.105 0.013699 0.604
GSE2197 167 1.197065779 5368 0.105 0.014286 0.604
GSE3982 169 1.201872999 1001 0.105 0.015625 0.604