Question about losing SNPs in result table

[。可乐不加糖]

Dear FUMA team,

I've been using FUMA for a long time.  Thank you for creating and maintaining such a convenient software!    

I encountered some issues while using it recently. I need to annotate significant SNPs derived from prior analyses. During my first submission with the required data format, many SNPs failed to be annotated. After switching to pre-define Lead SNPs, the number of missing SNPs decreased significantly, but several SNPs – like rs13190584 – have persistently remained unannotated. I've verified that this SNP exists in the 1000 Genomes European panel and is outside the MHC region. I'd like to understand why this might be happening. My job ID is 648772. Thank you very much.

Best regards,

Yao Zhao

[Tanya Phung]

Hi Yao, 

Is the SNP rs13190584 has the correct chr, pos, A1, and A2: 5:156493731:C:T?

I don't have access to the file that you submitted but it seems that you did not specify chr and pos but specified rsID. In this case, FUMA uses rsID to look up chr and position but it seems that A1 and/or A2 exists in your file. If that's the case and it does not match with the reference, the SNP would be filtered out. 

If you still have issues, you can send your input file to the FUMA email address (found on the website). 

Best,
Tanya

[。可乐不加糖]

Dear Tanya,

Thank you for your reply.

I've checked my input file, and the SNP  rs13190584 has the correct chr, pos  5:156493731 in pre-define lead SNPs file. But I did not  specify  A1, and A2.

Concrete input file were sent to the FUMA email address.

Thank you again for your time and effort.

Best regards,

Yao Zhao

[Tanya Phung]

Hi Yao,

When you submitted to FUMA using a file with just the rsID and p value, FUMA will then use dbSNP146 (processed by the original developer of FUMA). In the FUMA dbSNP146 database, this SNP is a biallelic variant: 
5       156493731       rs13190584      C       A,T

In this case because you did not further specify a1 and a2 allele, the variants that are biallelic are dropped from additional analyses. 
See: https://github.com/vufuma/FUMA-webapp/blob/master/scripts/gwas_file/gwas_file.py#L628

The solution would be to provide the file with the effect allele that matches the database. 

Best,
Tanya