Pre-defined lead SNP's not being outputted in Results

[Mihir Sanghvi]

Dear FUMA team,

Alongside uploading the summary statistics, I have uploaded a separate file in the "pre-defined lead SNP's option" (e.g. job number 245386). However, when the results are completed, zoom plots for the pre-specified variants are not available. I'd be most grateful for any advice.

Many thanks.

Best wishes,

Mihir

[Serafi Cambray]

Hi, 

I have the same problem. Predefined lead Snps are not considered,  It gives me the same results as when I did not put them. 

Regards, 

Serafí 

El dia dimecres, 12 d’abril de 2023 a les 15:49:45 UTC+2, Mihir Sanghvi va escriure:

[d.p.wi...@vu.nl]

Hello, 

Can you share your jobID so I can investigate further? If your job was deleted, can you please resubmit and then share the jobID with me?

Cheers,

Doug

[Anna Gatseva]

Hello everyone, 

I have the same problem. FUMA is only outputting one chromosomal region - even though from manhattan plot and  independent analysis I can see several other chromosomal locations that are < 5e-08. Also including list of lead snps doesn't include them in the results

Job ID: 273338

Any help will be appreciated

Kind regards,

Anna

[d.p.wi...@vu.nl]

Hi Anna,

This usually means that the other significant variants are not present in your chosen reference panel. I had a quick look at all of the significant variants in the manhattan plot and found that only 7 are in the UKB 10k EUR reference panel. It is possible that the alleles (or rsIDs) you provided don't match with 6 of them leaving only one signal on chromosome 15; however, I cannot check this because I cannot see your input file. I have included the 7 variants below:

15 31009544 15:31009544:A:G rs2067824
16 65188685 16:65188685:A:G rs74668049
17 6058036 17:6058036:A:G rs147815837
3 166796420 3:166796420:C:G rs1467252
5 80653626 5:80653626:A:T rs745427252
5 80782353 5:80782353:G:T rs116146760
5 80805670 5:80805670:C:T rs10050981

I would recommend using a different reference panel (like 1KG All) so fewer of your variants are lost, unfortunately that means your genomic risk loci will be defined based on LD estimates that may not match your study sample. If you have access to the raw genotype data which your sample is based on, you could use plink clumping using those LD estimates to create regions of interest around your significant variants. 

Cheers,

Doug

[Anna Gatseva]

Dear Doug,

That's very helpful - thank you so much for having a look

I do have access to the raw files so will try that and will have a go with the other reference panel

Kind regards,

Anna