No significant SNPs were identified.

[Siti Syazlinn]

Hello, i received this kind of error yesterday for my  job :

----------------------------------------------------------------------------------------------------------------------------------------------
ERROR: 5 (Error from lead SNPs and candidate SNPs identification / No significant SNPs were identified)
This error can occur when no candidate SNPs were identified. Note that indels are included in the FUMA from v1.3.0 but both alleles need to match exactly with selected reference panel. MHC region is also excluded by default. 1. If there is no significant hit at your defined P-value cutoff for lead SNPs and GWAS tagged SNPs, you can try to use a less stringent P-value threshold or provide predefined lead SNPs. 2. If there are significant SNPs with very low minor allele frequency, try decreasing MAF threshold (default 0.01). Manhattan plots and significant top 10 SNPs in your input file are available. 

----------------------------------------------------------------------------------------------------------------------------------------------

can i ask, does not having significant SNPs are a bad thing for SNPs association study?

[Andreas Schmidt]

How low is your "best" SNPs p-value?

[Siti Syazlinn]

actually, im not sure how to set the value of the significant snps p values.. mind to share? do u need my job id?

[Andreas Schmidt]

Sry, I am just a user like you, so I can't help you with your concrete data.

However, you asked if it is bad to have no "significant" SNP:

This indicates that none of your SNPs have a p-value of 5e-8 or lower.

There are now two possibilities with FUMA:

1. If your lowest SNP (check this in your data before FUMA) is between 1e-5 and 5e-8: Run fuma with a lowered threshold:

- Go to the section "2. Parameters for lead SNPs and candidate SNPs identification" when creating a fuma JOB

- Set "Maximum P-value of lead SNPs (<)" as high as possible

- Run the analyius

2. if the lowest p-value is higher than 1e-5:

Here it gets tricky! This means that you do not have strong association signals.

My personal experience:

- If your best p-values are between 5e-4 and 1e-5, there could still be some interesting signal:

You should define lead SNPs yourself (by using e.g. PLINK with 1000G data) and upload a list of lead SNPs when running fuma in the "Upload input files"-section here: "Pre-defined lead SNPs"

- However, if your best SNPs have p-values > 5e-4 you (should): rethink your hypothesis, check data formats, have a very small sample size and thus low power with your GWAS.

I hope this helps

Best,

Andreas