Hi everyone: I am going to run Metal to do a meta analysis.But now something make me confused.One data version is GRCh37 and the other one is GRCh38. Can Metal merge two GWAS Summary data even if the versions of SNP id are different?
Hi everyone: I am going to run Metal to do a meta analysis.But now something make me confused.One data version is GRCh37 and the other one is GRCh38. Can Metal merge two GWAS Summary data even if the versions of SNP id are different?
METAL by default will match variants by the MarkerName column, so you don't need chromosome or position information. However, as exemplified here, I strongly advise against this behavior as it can easily lead to matching variants that should not be matched and, vice versa, not matching variants that should be matched. A better approach is to use BCFtools/munge to convert summary statistics to GWAS-VCF files, use BCFtools/liftover to convert to GRCh38 all summary statistics not already on GRCh38, and then use BCFtools/metal, a METAL clone based on the GWAS-VCF specification, to perform the meta-analysis, all using the open BCFtools framework available here. This way variants will be matched according to chromosome, position, and alleles, minimizing the risk of spurious mismatches
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Thank you for your answer! I have another question: when I used liftover to convert GRCh38 to GRCh37, but the SNP id keep the same. Do I have to continue using the transformed coordinates to match to the SNP id of the GRCh37? If I don't need to do this, does it mean that the two versions even though the coordinates are different, but the SNP hasn't changed?
I randomly searched the SNP ids corresponding to several converted coordinates on NCBI and found that the snp ids did not change, so I am not sure whether the two versions only changed the location information while the SNP ids did not change.