Question about the methodology for selecting strong datasets

[harrsha4]

Hello,
Just as a followup to this thread, when you say the strong set was the lead eQTL for 20,000, do you mean within a tissue or amongst all tissues considering you were training on GTEx data? I am currently working with fastqtl results from 5 conditions (5 conditions x 33 million eQTL features) and used all lead SNP-eGenes from each condition (13,000 X 5 = 65,000) for the strong dataset with a 1 million random subset for the random data.

Thanks for creating this resource!

Harrsha

[pcarbo]

@harrsha4 In this vignette we suggesed one possibility: "Select the strong signals as those with lfsr < 0.05 in any condition in the 1by1 analysis."