observed shared genetic correlation a prerequisite for performing MR test

[Qingqin Li]

Could someone please comment on the point that if there is no shared genetic correlation at the genome level, there is no point to perform MR test (as it is inherently a whole genome method)? SNP A could be associated with two traits (as revealed by PheWAS), but the 2 traits may not be genetically correlated. Given the individual SNP may not be a strong instrument by itself, are we stuck here in figuring out whether SNP A has a pleiotropy effect on both traits, or SNP exerts its effect on trait Y via trait X?