Unable to run MTAG with --use_beta_se flag

[Mahantesh-Biradar]

Hello,

I'm trying to run MTAG with beta and SE, using the instructions from here:

Note: If both the effect sizes and the standard errors are present in the input GWAS summary statistics, z is not required by MTAG. To use this option, check that the repository is up-to-date and specify the flag --use_beta_se. The default column names are beta and se, which can be specified via the flags --beta_name and --se_name.

However, when I run MTAG using --use_beta_se option, I'm getting the following error. Can you please suggest how I can address this issue?

Traceback (most recent call last): File "/usr/local/bin/mtag", line 1546, in raise RuntimeError("Due to bugs in the beta-se code, this option has been temporarily removed from the MTAG software. Dec 2, 2021") RuntimeError: Due to bugs in the beta-se code, this option has been temporarily removed from the MTAG software. Dec 2, 2021

Thanks:)

[AudeIkuzwe]

I am having the same issue

[paturley]

A couple years ago, we noticed a few anomalies in MTAG results when the beta/se option was used that made us worry that the results weren't reliable. So we removed that option from the current release. I recommend you instead use the Z/N option (which is the default).

On Mon, Sep 11, 2023, 6:44 PM AudeIkuzwe @.***> wrote:

I am having the same issue

— Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/187#issuecomment-1714682717, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFBUB5PQYGBMKFF4BLWIZKTXZ6H5VANCNFSM6AAAAAA4UAADLM . You are receiving this because you are subscribed to this thread.Message ID: @.***>

[Mahantesh-Biradar]

Hi, thanks for letting me know about this. When we use the Z-score to run MTAG, do we need to do any post processing to MTAG_Beta values to make it comparable to beta values in input files? Thanks

[paturley]

I mentioned this in #185 as well. If you pass a single trait through MTAG, the beta values that are output should be a constant multiple of the input betas. The factor by which the betas and SEs are transformed is the same factor that they are in the multi-trait analysis.