MAGMA when no genome-wide loci

145 views
Skip to first unread message

Sander W. van der Laan

unread,
Nov 6, 2018, 10:48:05 AM11/6/18
to FUMA GWAS users
Hi,

When there are no GWAS hits - no genome-wide significant loci - it is great if MAGMA is run per standard. What would make it better, is if the actual MAGMA results file is also downloadable.

Thanks,

Sander

Kyoko Watanabe

unread,
Nov 6, 2018, 11:13:12 AM11/6/18
to FUMA GWAS users
Hi Sander,

Do you mean ERROR:005?
MAGMA still runs regardless of significant hits.
If there is no MAGMA results, either you de-activated MAGMA option or there was actually an error in your file to run MAGMA.
If you can share the jobID I'll have a look.

Best,
Kyoko

Xiangzhen Kong

unread,
Nov 6, 2018, 12:48:59 PM11/6/18
to FUMA GWAS users
Hi Kyoko, 

It seems that I have the same problem, i.e., ERROR:005. 
First, with this error, I coundn't found the tab for downloading the full MAGMA outputs. Yes, top 10 are shown under the tab Genome-wide plots. 
Second, it seems that MAF specified under 'Parameters for lead SNPs and candidate SNPs identification' did not work well. I tried several MAF, but the results still skipped several SNPs which showed low MAF. As I have filtered SNPs based on MAF, is it possible to set MAF as zero in FUMA? Thanks. 

Best,
Xiangzhen

Kyoko Watanabe

unread,
Nov 6, 2018, 3:12:19 PM11/6/18
to FUMA GWAS users
Hi Xiangzhen,

Sorry about the download, I'll fix that asap.
And yes, you can set MAF to 0, the MAF filtering is based on the MAF of the reference panel, so it's possible that MAF is much smaller in the reference panel than your actual GWAS samples.

Best,
Kyoko

amai...@gmail.com

unread,
Nov 7, 2018, 10:40:17 AM11/7/18
to FUMA GWAS users
Hi Kyoko,
I have also encountered that some SNPs are filtered out (i.e. don't show in annotations etc), even when MAF within my dataset for a given SNP is as high as 0.45.
I'm seeing this in job ID 26839 for SNP rs7420166, which looks fine for any other stats that I've checked in my dataset (e.g. HWE, INFO, MAF).
The SNP shows in the Manhattan plot, or even in locuszoom plots of neighbouring SNPs, but it's not annotated. It would be great to understand why it's been filtered out.
Thanks in advance!
amaia

Kyoko Watanabe

unread,
Nov 8, 2018, 11:39:14 AM11/8/18
to FUMA GWAS users
Hi Amaia,

I'm sorry for the delay.
I checked the SNP rs7420166 and it does not exist in out reference.
We used very strict filtering for UKB release 2 so it's possible that SNP was filtered out during our QC.
It does exist in release 1 but I will make release 2b (with additional imputed SNPs) available on FUMA.
I can't promise when it's going to be but I'll do my best to update by the end of this year.

Best,
Kyoko

Kyoko Watanabe

unread,
Nov 9, 2018, 3:45:51 AM11/9/18
to FUMA GWAS users
Hi all,

I have updated FUMA to be able to download MAGMA results for ERROR:005.
Below the top 10 SNPs table, you should be able to find an option to download.

Best,
Kyoko

amai...@gmail.com

unread,
Nov 12, 2018, 3:46:33 AM11/12/18
to FUMA GWAS users
Hi Kyoko,
Thank you for looking into it. Would it be possible to share the filtering parameters you used for the UKB release 2?
Thank you again for all your very useful work!
Best,
amaia

Kyoko Watanabe

unread,
Nov 12, 2018, 3:54:28 AM11/12/18
to FUMA GWAS users
Hi Amaia,

The details are described in the tutorial http://fuma.ctglab.nl/tutorial#refpanel
The process was performed by in-house UKB pipeline which we are not sharing the scripts.
But the same pipeline was used for the following two papers and they include all the details of the QC.

Best,
Kyoko

amai...@gmail.com

unread,
Nov 12, 2018, 10:02:50 AM11/12/18
to FUMA GWAS users
Great, thanks for all the info.
I think this SNP was excluded because they had been imputed based on the UK10K+1KG panel, not the HRC.
Best,
amaia
Reply all
Reply to author
Forward
0 new messages