FUMA currently supports hg19

[Tanya Phung]

Dear FUMA users, 

Currently FUMA only supports hg19. If your data is in GRCh38, you can:

1. use UCSC Liftover to obtain coordinates in hg19

2. provide your data with rsID (and not chr:pos in GRCh38). When only rsID is provided, FUMA looks up chr:pos from dbSNP version 146. 

Of course there will be some loss of variants due to the conversion not being perfect. We are working on updating FUMA resources to GRCh38 so thanks for your patience. 

Best,
Tanya

[Tanya Phung]

Dear FUMA users, 

As there have been some questions about the availability of GRCh38, this is the update: 

• Because most of the reference databases used in the SNP2GENE and GENE2FUNC modules are in GRCh37 coordinates, it makes most sense if your GWAS sumstat is in GRCh38 to lift down to GRCh37. I may implement an automatic liftover if time allows. 
• However, lift over can result in loss of variants so this solution is not ideal. Ideally resources utilized by FUMA are in GRCh38 coordinates. As this involves reprocessing of (lots of) data, this will take some time to be worked on. 

In summary: 

• If your GWAS sumstat is in GRCh38 and you want to use FUMA, loss of variants is unavoidable. You can use either (1) use UCSC Liftover to obtain coordinates in GRCh37 or (2) provide your data with rsID (and not chr:pos in GRCh38). When only rsID is provided, FUMA looks up chr:pos from dbSNP version 146.
• I'll plan on updating the reference database that FUMA utilizes to GRCh38 but there is no predicted timeframe of when this will be available. 

Best,
Tanya

[Sagittarius]

hi

when i use the only rsID (build 38), I get the manhattan plot but no result on eQTL or gene-based results.

basically when i upload sum-stat, i only call the box for rsID  (as SNP) and blank the chr and position.

is this correct approach? please let me know

thanks

[Tanya Phung]

Hi, 

What is your jobID? 

Tanya

[Sagittarius]

HI

the ID is  738177

thanks

[Tanya Phung]

Hi, 

Based on this manhattan plot from your job, none of the variants are genome-wide significant: 

The dotted red line denotes the genome-wide significant threshold, which is equal to -log10(5e-8) = 7.3. None of the variants in the submitted gwas sumstat passes this threshold. Therefore, additional analyses were not carried out. 

Best,
Tanya

[Sagittarius]

I understand...however i set a minimum p value of 10-5  (1e-5) in the start up page under Maximum P-value of lead SNPs (<)   . I usually see some results with this value. I think the issue might be related to something else.

[Tanya Phung]

Hello, 

Ok gotcha. There should be some variants where the p values < log10(5e-5) or <4.3. 

The problem is that in your gwas sumstat file that you submitted to FUMA, the columns for chr and/or position (which are in GRCh38) are still present. In the case that only the rsID is present, then, it will print in the log file: "Either chr or pos is not provided". However, this was not printed out in the log.

For this functionality to work, you need to remove the columns for chromosome and position before submitting. You can also share the first few lines of your file so I can check. 

Best,
Tanya

[Sagittarius]

great! thank you!