Selecting SNPs for two-sample Mendelian randomization study/analysis

Question

[huynguyenacademia]

Dear All,

I'm familiar with one-sample MR analysis. However, when it comes to two-sample MR, I feel a bit mixed-up of how to choose SNPs and prepare data ready for analysis. To simplify, suppose I have chosen 11 SNPs for exposure (X) and 16 SNPs for outcome (Y); however, almost or even all SNPs for exposure are not SNPs for outcome (see a hypothetical example ).

So, to prepare effect size (beta) and SE for each set of SNPs, they are not overlapping SNPs, so cannot perform two-sample MR analysis, while we are expecting the standard layout for analysis as this: .

So, I appreciate any of your advice or guide on how to choose SNPs for 2-sample MR.