Decimal places in EigenStrat p-values?

Hello,

I have successfully managed to run EigenStrat (smartpca and smarteigenstrat) on a collection of ~34k SNPs (I am aware of the potential loss of power). I ran smartpca on all of the 34k SNPs (which have been filtered from a larger SNP dataset to remove linkage disequilibrium, and a MAF 0.05 filter applied), and then tested all 34k SNPs for association.

When looking at the output, however, I note that p-values only down to 4 decimal places are reported, with many loci having reported values of '0.000' in my .chisq output file (regardless of whether I use 10 or 30 PCs, although I wouldn't expect this to make a difference).

Is there any reason why the p-values seem to go no lower than 4 decimal places? I find it hard to believe that are that many significant loci in my dataset, especially since other methods I have tried find very few if any, and the threshold for a simple Bonerroni correction I wish to use at this point is 10^-5.

I have read over the documentation several times, but cannot find any mention of this issue/any options concerning this output, or why it might be occurring.

I feel as though I must be overlooking something, or perhaps not running EigenStrat correctly, although I had no trouble running smartpca.perl or smarteigenstrat.perl. Note that I used ped/map files as input.

Any help in understanding this issue is much appreciated!

Thank you,
Conrad Izydorczyk

Start by reproducing the smartpca and smarteigenstrat run with the stated ped/map inputs and inspect the .chisq output precision. Determine whether the reported 0.000 values are formatting or calculation results; done means explaining the behavior or reporting p-values accurately enough to evaluate a 10^-5 threshold.