Comparison of pleiotropy across populations

[explodecomputer]

With Maxz instruments, we think we've minimised heterogeneity acrosss studies that could be due to mismatches in LD.

So if there are outliers in population 1, that is because of pleiotropy. Are these SNPs also outliers in population 2?

Hypothesis - we can predict the outliers in pop2 based on outliers in pop1 because horizontal pleiotropy is conserved across populations

To do this

for pop 1. get the per-snp q statistics using radialMR do the same for pop2.

Does outlier status in pop1 predict outlier status in pop2? I think just a chisq test

[yoonsucho]

Method: 1) Identify the outliers in each population (Radial MR) 2) Obtain outlier effect using MR-TRYX 3) Run estimate_instrument_specificity() to see whether outlying SNPs in pop1 are also identified as outliers in pop2

Questions: 1) outlier effect: Original ratio (b.outcome / b.exp) - adjusted ratio (adjusted.b.outcome/b.exp)? 2) How to calculate se for the outlier effect: Use delta method? https://stats.stackexchange.com/questions/19576/variance-of-the-reciprocal-ii/19580#19580 https://www.researchgate.net/post/How_do_you_calculate_the_standard_deviations_when_subtracting_two_means_with_different_standard_deviations