Losing two lead SNP in result table

[程畅]

Dear FUMA team, 

I have a recent job ( 559494 ). There are 11 signals in the SNP Manhattan plot. But there are 12 lead SNP summarized in the result table. Two SNPs on chromosome 6 were missing from the results,and one and two SNPs on chromosomes 3 and 10 were extra, respectively.

This happened for my multiple jobs. 

Could you help me understand why this happens and how to solve this? 

Thank you very much

The following is the specific information of the lead  SNPs  I found through the Manhattan plot in my GWAS data.

[Tanya Phung]

Hello, 

1. 2 SNPs on chromosome 6 were missing from the results
   1. SNP: 6:29973925:C:T: this SNP falls within the MHC region (chr6 between positions 29614758 and 33170276. When you submitted your GWAS summary statistics, you selected to exclude MHC region. If you want to include the SNPs that fall within the MHC region, you need to UNCHECK this option in section 5 
   2. SNP: 6:82969138:T:A: This SNP does not exist in the reference database that you selected, which is UKB release2b EUR_10k. In fact, I checked other resources such as gnomAD and this variant does not exist either. There is nothing we can do if the variant does not exist in the database.
2. For the one SNP on chromosome 3 and two SNPs on chromosome 10 you mentioned that are extra, did you double-check that they exist in your input file?
   1.   We do not have access to your input file - they are deleted as soon as the job is finished. Therefore, I can not check if these 3 SNPs exist in your input data. However, from the file that plotted the manhattan plot, I can see that these 3 SNPs exist:

less manhattan.txt | grep 190675956 3 190675956 3.83874e-08 less manhattan.txt | grep 104272336 10 104272336 4.9456e-10 less manhattan.txt | grep 105628548 10 105628548 1.61147e-08  

Therefore, it appears to me that these 3 SNPs do probably exist in your file. Please check your file to make sure.

[chang cheng]

  Thank you very much for your response; it has solved a major issue for me. Additionally, regarding the second question, I meant that on the Manhattan plot, chromosomes 3 and 10 show only one lead SNP each. However, FUMA identified more than one, and the additional SNPs are located very close to the most significant SNP. These should have been filtered out through LD pruning, but they were not. Could you please advise on how to resolve this issue?  

[Tanya Phung]

Hello, 

According to the FUMA tutorial: https://fuma.ctglab.nl/tutorial#snp2gene

FUMA uses r2 threshold of 0.6 to define independent SNP.

On chr3, these 2 SNPS:
3:190650833:A:G
3:190675956:A:C

have an r2 value of '0.0728839 in the UKB release2b EUR_10k reference panel. Because r2<0.6, they are independent. 

The 3 SNPs on chr10 are all independent. You can check this using an LD calculator, though you won't be able to select the same reference population but you can select a EUR population to check. 

10:104272336:C:T

10:105459116:A:G

10:105628548:G:GA

Best,
Tanya