Not sure what do you mean by sign test, but why don't not synthesize (meta-analyze) the effect sizes of these datasets to increase your power?

Thank you Evangelos Evangelou. Sign test in context of GWAS means if the same allele has similar or opposite effect in two populations. For example if Allele A increases the risk in one population and it increases the risk in the second population as well then we can assume that with bigger sample size the p value will become significant in the second population. I take your point about meta-analysis. PGC2 results are from meta-analysis of studies. Our sample size is very small compared with PGC2 sample size and its ancestary is different from PGC2 samples. Sign test can give us some idea if further collection from this different population will produce results which can be combined with PGC2 results.

Dear Muhammad, 

Normally sign test is calculated as a non-parametric test with n less than 25 matched pairs. The minimum number pairs is 6. You have two columns of numbers side by side and check how many minus or plus you get (less or greater than) . Then you calculate your signs + or -  and look up in a table if your number of signs is significant. . For larger samples see Sidney Siegel. Nonparametric statistics för behavioral sciences, International student edition McGraw-Hill Kogakusha LTD.

A template for the sign test using Excel is available at http://www.real-statistics.com/non-parametric-tests/sign-test/

Check also

https://en.wikipedia.org/wiki/Sign_test