Is there a list of high Linkage Disequilibrium (LD) regions for genome Build 38?

Question

[yy237]

I am conducting GWAS QC and would like to exclude regions of high linkage disequilibrium (LD) during pruning of my data (which is in GRCh38). Does anyone know where I may find a list of high LD positions in build 38? ... or how may I map the positions listed in https://genome.sph.umich.edu/wiki/Regions_of_high_linkage_disequilibrium_(LD) to build 38?

I also found a list in build 37 in De Vlaming et al., 2017 (Table 5) (https://www.biorxiv.org/content/10.1101/211821v1 ), but am not sure what would be the best approach.

Thank you.

EDIT: I tried to use liftOver to map the list of high LD regions from De Vlaming et al., 2017 to Build 38 using the commands below, but there were plenty of unlifted positions in both (in brackets are the row counts in the output and unlifted bed files).

liftover input.bed hg18ToHg19.over.chain.gz outputA.bed unliftedA.bed

(16 positions lifted, 18 positions unlifted)

liftover input.bed GRCh37_to_GRCh38.chain outputB.bed unliftedB.bed

(15 positions lifted, 20 positions unlifted)

Does liftOver prefer hg18ToHg19.over.chain.gz or GRCh37_to_GRCh38.chain?

Would it be okay for me to use the output.bed files even though many of the high LD regions were not lifted from the original b37 list?

Thank you.

Answer 1

[yy237]

Hi, I ended up using the NCBI Remap tool (https://www.ncbi.nlm.nih.gov/genome/tools/remap/ ) to remap the list of positions I have. Since I am unsure of which release versions of genome builds to use, I tested out the following combinations (1) GRCh37 (hg18) to GRCh38 (hg19) (2) GRCh37 (hg 18) to GRCh38p13 (release 13) (3) GRCh3p13 (release 13) to GRCh38

I noticed that regardless of GRCh38 (hg19) or GRCh38p13 (release 13), the positions of my mapped positions (output) were the same). As for comparing GRC37 (hg18) versus GRCh37p13 (release 13), I noticed the second option above (in bold) gave me an extra mapped position, and think I should proceed with the output from that. Can someone advise if this approach makes sense?

Thank you.