Great thanks for this beautiful package and all the efforts in harmonising the GWAS summary data as well integrating the MR methods.
I have recently noticed there are some bugs with the harmonised GWAS data; e.g the GWAS summary data for the UKBB biomarker data released by Neale Lab. For example, the sample size in the database for HDL GWAS (id = ukb-d-30760_irnt) was set to NA but this was reported in the original GWAS summary data, also the P-values for some variants were wrongly reported (e.g. for rs12720922, the P-value in the original summary data was P=0, but in the TwoSampleMR package it was reported as P=1), P= 0 or 1 is hugely different and may lead to really wrong conclusions. Of course, we could estimate the correct P using beta and se, but I think at least I would let you know these bugs.
Dear Developing Team
Great thanks for this beautiful package and all the efforts in harmonising the GWAS summary data as well integrating the MR methods.
I have recently noticed there are some bugs with the harmonised GWAS data; e.g the GWAS summary data for the UKBB biomarker data released by Neale Lab. For example, the sample size in the database for HDL GWAS (id = ukb-d-30760_irnt) was set to NA but this was reported in the original GWAS summary data, also the P-values for some variants were wrongly reported (e.g. for rs12720922, the P-value in the original summary data was P=0, but in the TwoSampleMR package it was reported as P=1), P= 0 or 1 is hugely different and may lead to really wrong conclusions. Of course, we could estimate the correct P using beta and se, but I think at least I would let you know these bugs.
Best wishes, Qin