Continuous MAGMA ERROR

[Alero Okoh]

Hello FUMA Team,

Kindly assist me in rectifying the continuous error (ERROR:magma) I get when attempting to run MAGMA analysis on data with jobID 487114

Kind regards

 

[Alero Okoh]

I have also tried to run the analysis without ticking the MAGMA section. I get a NaN in the Manhattan plot. How can I rectify this error?

[Tanya Phung]

Dear Alero, 

The job  487114 no longer exists on FUMA server so we can't check what went wrong. For your second issue, what's the jobID? 

However, based on the description, it sounds like the pvalue column was perhaps incorrect? 

Best,
Tanya

[Alero Okoh]

Dear Tanya,

Thanks for your feedback. The Job IDs are  487111 and  487399. Kindly give your assessment of what the problem might be so I can correct them and retry the analysis.

Thank you very much

Ofure

[Alero Okoh]

Dear Tanya,

In my new job with ID  487541 which contains all my SNPs of interest, I unchecked the MAGMA box and got an output. However, I still get a NaN in my Manhattan plot and my QQ plot appears off the chart. These issues are similar to the ones I mentioned above, kindly assist me with ways I could rectify the errors. I would have liked to run a MAGMA

Warm regards,

Ofure

[Tanya Phung]

For job 487111, this is the magma error:
read 23 lines from file, containing valid SNP p-values for 16 SNPs in data (69.57% of lines, 6.384e-05% of SNPs in data)
Loading gene annotation from file /data/MAGMA/1KG/Phase3/EUR_ENSGv102_w0.genes.annot...

ERROR - reading gene annotation file: found no genes containing SNPs in genotype data

Terminating program.

Please refer to MAGMA manual for questions related to issues with MAGMA: https://cncr.nl/research/magma/

For job 487399: in this one, magma was unchecked. What's the exact problem?

For job  487541, could you send the picture of the manhattan plot?
For the QQ plot, it seems that your input file has 24 variants. FUMA is designed to analyze GWAS (genome-wide data) so a qq plot from 24 variants might not be interpretable.

Best,
Tanya

[Alero Okoh]

Okay Tanya, I have noted your feedback. Kindly see the attached file for the Manhattan plot.

[Tanya Phung]

Hi, 

FUMA is not designed to analyze a selection of highly significant p-values. 

The qq plots the quantiles of the expected log p vs the observed log p - ‘the expected’ is the distribution of p values under the null hyp of no association. If the observed is a selection of highly significant p-values, ofc the qq plot will look weird. 

You should upload the full sumstats to interpret the qq-plot in terms of indication of population stratification of any other biases. 

Thank you,
Tanya

[Alero Okoh]

Dear Tanya,

Thank you very much for all the clarification provided, it's much appreciated.

Warm regards,

Ofure