Liftover of SNP genotype calls GrCH387 to GRCH38 - why not just use rsids as recommended by UCSC?

Gabriel Doctor

Hi,

The UCSC specifically advises against using LiftOver for SNP genotype data and instead urges researchers to use rsids where available: https://genome.ucsc.edu/FAQ/FAQreleases.html#snpConversion

Your documentation points to a fairly complex liftover pipeline whcih does just what is recommended against. 

Why not just provide a list of corresponding rsids with correspond loci in build 37 and 38 (and falling back on  lift over/chrom-pos a coordinates for Affymetrix ids)? 

This would be trivial to create using the UCSC table browser.   Does this exist already? If not, why is there a reason this simple solution not been implemented? 

Comments

2 comments

  • Comment author
    Gabriel Doctor

    It would be a small file with 3 columns! ID, position_GRCH37, position_GRCH38, approx 800k rows.  Simple to update a plink.bim file with this and much cheaper for users than the liftover pipeline

     

     

    0
  • Comment author
    Rachael W The helpers that keep the community running smoothly. UKB Community team Data Analyst

    Hi Gabriel,

    thank you for the suggestion, which will be passed to our bioinformaticians for consideration.   In the meantime, for those SNPs that have an rsid, you might find the dbSNP database useful.   For example, for rs799917 see this page of the NCBI website https://www.ncbi.nlm.nih.gov/snp/?term=rs799917 .

    Thank you for using the forum.

    0

Please sign in to leave a comment.