Troubleshooting Job ID 558783, ERROR:001

[Lauren Spor]

Hi there,

This is my first time using FUMA and I have many failed attempts at formatting the data to run this program. Could you help me understand what is wrong with my input file?

I have:

• converted my genome from hg38 to hg19
• removed rows that were incorrectly mapped to chrY during liftover
• submitted a tab-deliminated txt.gz file (zipped with gzip)
• correctly indicated the column names for chromosome, position, reference allele, alternate allele, p-value, and number of subjects during submission
  • I did not include any rsIDs
• chromosomes are integers ranging from 1-23
• positions are numeric whole numbers, not in scientific notation
• confirmed that none of the rows are missing data for chromosome, position, or p-value
  • some alleles are missing. Is this an issue? The webpages I looked at previously said this would be okay
• confirmed that there are equal numbers of columns and column names. The header does not begin with "#"
• confirmed there are no quotation marks in the file
• confirmed that column names do not have extra whitespace. I checked the first 10 lines of my file and the only white space are the tab deliminations 

Would you please advise me on what part of my formatting needs to be changed? Additionally, it would be helpful to show a snippet of an example input file for submission on the tutorial page for people to model their submission after. Thanks!

Lauren

[Tanya Phung]

Hi Lauren, 

This is the error of your job: 
invalid literal for long() with base 10: '1.45e+08'

Therefore, it looks like for some positions, they are still in scientific notation. 

This is the first few lines of the example file. 

rsID    chr     pos     ref     alt     p
rs3934834       1       1005806 C       T       0.71
rs3766192       1       1017197 C       T       0.61
rs3766191       1       1017587 C       T       0.8
rs9442372       1       1018704 A       G       0.59
rs10907177      1       1021346 A       G       0.78
rs3737728       1       1021415 A       G       0.47
rs9442398       1       1021695 A       G       0.28
rs6687776       1       1030565 C       T       0.1
rs6678318       1       1030633 G       A       0.11

Best,
Tanya

[Lauren Spor]

Thanks Tanya,

I fixed the issue and successfully ran FUMA with job ID 559375, however, I forgot to select MAGMA.

My repeat run (job ID 559388) could not run MAGMA (ERROR: magma). I kept the default settings when I ran it (see screenshot below).

I then ran it one more time, but selected a window of "1," but got the same ERROR: magma (job id: 559427). What is going wrong?

Best,
Lauren

[Tanya Phung]

Hi Lauren, 

I checked the parameters that you selected, and I saw that for the sample size column name, you put in Weight. I can't see your input GWAS summary statistics (it is automatically deleted after a job), so I cannot check if this column does indeed contain the sample size. Could you check that? 

Changing the gene window size is not relevant here. The gene window indicates how big the region upstream and downstream the gene boundary is.

You can also send us your input file to fuma....@gmail.com if you still have issues. 

Best,
Tanya

[Lauren Spor]

Hi Tanya,

Yes, the weight is the sample size from the METAL output. The number of samples varies a bit from SNP to SNP. Should I use the max number of participants instead of varying it by SNP?

Other potential issues that come to mind are whether the p-value must also be an integer, rather than scientific notation, or if the alleles cannot have missing data. The Weight column is in integer format. Thoughts?

Best,
Lauren

[Tanya Phung]

Hi Lauren, 

FUMA supports both option N and varying N by SNP, so that's not the issue. 

P value can't be integer (unless it's 1), and FUMA accepts scientific notation for P values. 

I suggest that you send your input file to the fuma email address as above, and we can check it. 

Thank you,
Tanya

[Lauren Spor]

Thanks Tanya,

I just emailed you a file containing 20% of the SNPs. The full file was too large to send.

[Tanya Phung]

Hi Lauren, 

Thank you for sending the file. 

The issue is: you specify that alt is the eacol and ref is the neacol when you submitted FUMA jobs. However, for some rows, these information are missing. For example, from the file that you sent:

When there are missing data like this, for some part of the pipeline, FUMA is unable to determine the correct column for the p value. Therefore, in this case, MAGMA failed. 

There could be 2 options: 

1. You can remove all rows with missing data 

or

2. You can remove the alt and ref columns. Then, FUMA will fill in these columns (but it will be slower to run). However, you still need to make sure that there are no rows where the p values are missing. 

I hope this solves your issue. 

Best,
Tanya

[Lauren Spor]

Thanks Tanya!

My analyses are running now that I removed rows with missing alleles.