Question about opposite BETA directions in MTAG results when changing input file order

[shaoyu-nanjing]

Hello!

I have a question regarding the BETA directions in MTAG results when the order of input summary statistics files is swapped. Specifically:

I noticed that when I run MTAG with two traits (--sumstats A,B vs. --sumstats B,A), the BETA values for the same SNPs in the output files appear to have opposite signs.
After inspecting the results, it seems the direction of the BETA values is determined by the first input file. For example, when A is the first input file, the BETA values are consistent with the direction of effect in A. However, when B is the first input file, the BETA values appear to reverse their signs.

And I found this change does not affect the values, so I want to confirm:

Is this behavior expected in MTAG?
Does this sign change reflect a difference in the analysis, or is it solely a matter of how MTAG defines the effect allele direction based on the primary input file?
Are there any best practices to standardize the output direction for easier interpretation and downstream analysis?

Thank you for your help!

[paturley]

Hello, Yes, MTAG uses the effect allele from which ever trait is listed as trait one. MTAG should report the correct reference allele corresponding to the output results, so as long as the effect allele is taken into account, it should have no effect on downstream analyses. Best, Patrick

…

On Tue, Nov 19, 2024 at 8:05 AM shaoyu-nanjing ***@***.***> wrote: Hello! I have a question regarding the BETA directions in MTAG results when the order of input summary statistics files is swapped. Specifically: I noticed that when I run MTAG with two traits (--sumstats A,B vs. --sumstats B,A), the BETA values for the same SNPs in the output files appear to have opposite signs. After inspecting the results, it seems the direction of the BETA values is determined by the first input file. For example, when A is the first input file, the BETA values are consistent with the direction of effect in A. However, when B is the first input file, the BETA values appear to reverse their signs. And I found this change does not affect the values, so I want to confirm: Is this behavior expected in MTAG? Does this sign change reflect a difference in the analysis, or is it solely a matter of how MTAG defines the effect allele direction based on the primary input file? Are there any best practices to standardize the output direction for easier interpretation and downstream analysis? Thank you for your help! — Reply to this email directly, view it on GitHub <#222>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFBUB5PYZKOCKPPBGITD52D2BMZQ7AVCNFSM6AAAAABSCCC6LKVHI2DSMVQWIX3LMV43ASLTON2WKOZSGY3TEMBZG44DCOI> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[shaoyu-nanjing]

Thank you very much for your response! Actually we still have some additional questions:

When applying the --equal_h2 parameter, we encountered the following issue: in some cases, both the BETA values for trait1 and trait2 are positive, but after meta-analysis, they both turn negative. We are unsure if there is a specific reason to explain this or if we have misapplied this parameter. This might be a silly question, but it has indeed left us quite confused.

Thank you again for your help!

[paturley]

Have you verified that the effect allele is the same?

…

On Fri, Nov 22, 2024, 8:34 AM shaoyu-nanjing ***@***.***> wrote: Thank you very much for your response! Actually we still have some additional questions: When applying the --equal_h2 parameter, we encountered the following issue: in some cases, both the BETA values for trait1 and trait2 are positive, but after meta-analysis, they both turn negative. We are unsure if there is a specific reason to explain this or if we have misapplied this parameter. This might be a silly question, but it has indeed left us quite confused. Thank you again for your help! — Reply to this email directly, view it on GitHub <#222 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFBUB5KTWFGW5LDTUI7RSBD2B4XH5AVCNFSM6AAAAABSCCC6LKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIOJTG44DENJQGA> . You are receiving this because you commented.Message ID: ***@***.***>