Issue running SNP2GENE jobs on FUMA platform

[Olivier Robert]

Dear Ms. Phung,

I am writing to report a persistent issue I have been experiencing with the SNP2GENE function on the FUMA platform since March 24.

Since that date, I have been unable to successfully run any jobs. Each attempt fails with the message «ERROR: timeout». Ex: with job #724211

To troubleshoot, I have:

• Tried running multiple jobs at different times
• Used relatively small input files
• Re-ran previous datasets that had successfully worked in the past using the same submission conditions and analysis parameters

Despite these efforts, all jobs consistently fail with the same timeout error.

Additionally, my supervisor has also attempted to run older datasets that previously worked, and encountered the same issue.

Given that this problem has persisted for several days and affects different datasets and users, I would greatly appreciate your assistance in understanding the cause and identifying a possible solution.

Please let me know if you need any additional information (e.g., job IDs, input files, parameters).

Thank you very much for your help.

Best regards,

Olivier Robert

[Tanya Phung]

Hi Olivier, 

For the job 724211 it was terminated because it was taking more than 8 hours. There is a limit of 8 hours per job. 

Even though the input gwas summary statistics file that you submitted was deleted from the FUMA server, from an intermediate file, I gather that there are at least 1838978 variants in your file and it seemed that all of these are significant (the highest p values of these variants is 1e-09 which is lower than the genome-wide threshold of 5e-8). What this means is that all of these ~2 million snps are being analyzed. This is not what we expect from a GWAS. To put it in perspective, in a recent AD GWAS, there are approximately 7.5 million snps in the input file but only around 20k snps are genome-wide significant. In summary, analyzing a file with close to 2 million significant snps will take more than the allocated 8 hours. 

If you want me to look at the jobs where you used smaller input file and/or attempting to repeat previously ran analyses, please send the jobIDs (where the jobs have not been deleted). 

Best,
Tanya