Predefined lead SNPs do no match lead SNPs calculted by FUMA

[willemijn...@gmail.com]

Hi,

I have a question. I have used both FUMA and a manual way to determine lead SNPs, but they don't give the same results.

My command for the manual way is:

/hpc/local/CentOS7/dhl_ec/software/plink_v1.9 --bfile /hpc/local/CentOS7/dhl_ec/software/MetaGWASToolKit/RESOURCES/1000Gp3v5_EUR/1000Gp3v5.20130502.EUR.noDup  --clump /hpc/dhl_ec/wopdenbrouw/UKB/GWAS/Clumping/GWAS_pre.txt --clump-snp-field "RSID" --clump-p1 5e-8 --clump-p2 0.005 --clump-r2 0.05 --clump-kb 500kb --clump-field "P" --out /hpc/dhl_ec/wopdenbrouw/UKB/GWAS/Clumping/Prediabetes_clumping.clumped --clump-verbose --clump-annotate CodedAlleleB,OtherAlleleA,CAF,MAF,MAC,HWE,AvgMaxPostCall,Info,BETA,SE.

I used the same setting in FUMA and I even tried uploading these pre-defined lead SNPs in FUMA. However, I keep on getting different results. FUMA gives me around 200 lead SNPs, while the command above gives me around 280 lead SNPs. Also, only around 80 of these SNPs overlap.

I hope someone can help me with this.

Willemijn op den Brouw

[Kyoko Watanabe]

Hi Willemijn,

One of the differences is coming from how r2 was calculated.

Please refer to this for more details.

https://groups.google.com/d/msg/fuma-gwas-users/ZkuLqf2pS0A/9CQZs07pBQAJ

In addition, there are multiple reasons why the lead SNPs are different.

1. SNPs in the reference panel is not the same, so some of your lead SNPs might not exist in the reference panel on FUMA

2. FUMA uses unique ID consisting of chromosome, position and alleles rather than rsID since rsID is not unique. Since you are using rsID in your plink command, that might cause some differences

3. SNPs might be filtered due to MAF in FUMA

If you can share your SNP2GENE jobID and your pre-defined SNPs, I will be able to give you more specific answer.

Best,

Kyoko