Missing Lead SNPs

Sarah Benstock

unread,

Jun 26, 2023, 10:08:58 PM6/26/23

to FUMA GWAS users

Hi,

I am having a problem where I have a genome-wide significant hit on a chromosome, however, the lead SNP is not present in any of the results (Lead SNP table, chromatin interactions etc.). I am having this issue on 3 different jobs for the same locus on Chr6 (Job IDs are below). I'm not sure where the problem is... any help would be appreciated!

Thanks,

Sarah

Job IDs

263105

263104

263102

Marijn Schipper

unread,

Jun 27, 2023, 4:16:32 AM6/27/23

to FUMA GWAS users

Hi Sarah,

Are the SNPs that are missing part of the reference panel you selected? You can find more info on that here: https://fuma.ctglab.nl/tutorial#refpanel.

As FUMA is a sumstats only method, we need LD information from a reference panel in order to be able to do LD based clumping of SNPs into loci.
FUMA therefore drops SNPs which are not part of your selected reference panel.

Please let me know if your SNPs are indeed part of the reference panel you selected, I'm happy to help you troubleshoot.

Cheers,
Marijn

Op dinsdag 27 juni 2023 om 04:08:58 UTC+2 schreef Sarah Benstock:

Sarah Benstock

unread,

Jun 27, 2023, 11:33:37 AM6/27/23

to FUMA GWAS users

Hi Marijn,

I have double checked, and the SNPs are definitely part of the reference panel. If it's helpful the rsID is rs3131113, and the MAF is 0.312, so it's not a rare variant either.

Any further help would be greatly appreciated!

Thanks,

Sarah

Marijn Schipper

unread,

Jun 27, 2023, 4:11:43 PM6/27/23

to FUMA GWAS users

Hi Sarah,

Good to know!
Could you send a part of the input sumstats that you used that contains any of these disappearing SNPs to m.j.sc...@vu.nl?
I'll take a look to see what's going on.

Cheers,

Marijn

Op dinsdag 27 juni 2023 om 17:33:37 UTC+2 schreef Sarah Benstock:

Marijn Schipper

unread,

Jun 28, 2023, 2:36:46 AM6/28/23

to FUMA GWAS users

Hi Sarah,

I just checked your SNP, and it seems that this SNP falls in the MHC region. We drop MHC SNPs since using reference panels for LD estimations isn't very accurate. If you still want this SNP included you can rerun your analysis and deselect the exclude MHC region in section 5 of the input page when uploading your sumstats.

Hope this helps!

Cheers,
Marijn