How to choose or how to explain the result

[caozilongsuper]

Hi, MTAG programer, I have got my analysis result successfuly, but I am confused about the final result. I combined two subtypes of disease which used different cases and same controls. So next which file is the combined file I can use in following analysis. Results such as : Trait # SNPs used N (max) N (mean) GWAS mean chi^2 MTAG mean chi^2 GWAS equiv. (max) N 1 finngen_bc.txt 14706486 306119 306119 1.066 1.067 310396
2 finngen_sc.txt 14706486 290388 290388 1.007 1.066 2701022

[paturley]

It depends on what you are trying to do. MTAG is meant to produce summary statistics that correspond to each of the input phenotypes.

However, from this summary table, I'm noticing that there is a pretty large power difference between the two input GWASs, which may lead to a higher FDR for your lower powered trait. Not necessarily a problem if you are confident the two traits are highly correlated, but just a risk.

On Mon, Nov 6, 2023 at 9:00 AM caozilongsupper @.***> wrote:

Hi, MTAG programer, I have got my analysis result successfuly, but I am confused about the final result. I combined two subtypes of disease which used different cases and same controls. So next which file is the combined file I can use in following analysis. Results such as : Trait # SNPs used N (max) N (mean) GWAS mean chi^2 MTAG mean chi^2 GWAS equiv. (max) N 1 finngen_bc.txt 14706486 306119 306119 1.066 1.067 310396 2 finngen_sc.txt 14706486 290388 290388 1.007 1.066 2701022

— Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/193, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFBUB5OUIZ4XISLRK7XS2STYDDUQLAVCNFSM6AAAAAA67R3DROVHI2DSMVQWIX3LMV43ASLTON2WKOZRHE3TSMRSHA3TGOI . You are receiving this because you are subscribed to this thread.Message ID: @.***>

[caozilongsuper]

Thanks for quick reply! I am wondering what meaning of "GWAS equiv. (max) N" is. And what's the final sample number I can adapt.The main purpose is combining the two subtypes diseases as one.What the difference between two "trait1,trait2" files? Another question is colum called "mtag_beta" is also different cross two result "trait" files when comparing same rsid, which one should I use as the final value to measure this rs effect? I hope my confusion not bother you and thanks a lot!

[paturley]

This is explained in the MTAG paper. The formula for it is N_GWAS*(1-meanChi2_MTAG)/(1-meanChi2_GWAS). It's meant to approximate how large a sample you'd need to produce a GWAS that equally powered to the MTAG results for that phenotype.

On Tue, Nov 7, 2023 at 4:00 AM caozilongsupper @.***> wrote:

Thanks for quick reply! I am wondering what meaning of "GWAS equiv. (max) N" is.

— Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/193#issuecomment-1798075570, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFBUB5JFBTMPUQ6F7GX4ZQLYDH2C3AVCNFSM6AAAAAA67R3DROVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTOOJYGA3TKNJXGA . You are receiving this because you commented.Message ID: @.***>

[caozilongsuper]

Thanks for your explanation and I got it! How about next question mentioned above?

[paturley]

Which question? Sorry, I'm not sure what you are referring to.

On Tue, Nov 7, 2023 at 11:04 AM caozilongsupper @.***> wrote:

Thanks for your explanation and I got it! How about next question mentioned above?

— Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/193#issuecomment-1799035727, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFBUB5PZX6ORQY66MN5SPSLYDJL2TAVCNFSM6AAAAAA67R3DROVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTOOJZGAZTKNZSG4 . You are receiving this because you commented.Message ID: @.***>

[caozilongsuper]

Another question is what the difference between two "trait1,trait2" files? And colum called "mtag_beta" is also different cross two result "trait" files when comparing same rsid, which one should I use as the final value to measure this rs effect? I hope my confusion not bother you and thanks a lot!

[paturley]

No worries. As I mentioned above, MTAG produces one set of summary statistics for each trait. So the difference in the effect size estimates is because the estimated effect of the SNP on each trait is different. The set of summary statistics you should use depends on which trait you want summary statistics for.

On Wed, Nov 8, 2023, 4:02 AM caozilongsupper @.***> wrote:

Another question is what the difference between two "trait1,trait2" files? And colum called "mtag_beta" is also different cross two result "trait" files when comparing same rsid, which one should I use as the final value to measure this rs effect? I hope my confusion not bother you and thanks a lot!

— Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/193#issuecomment-1801360900, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFBUB5L6KMBNCW3663Q5LKLYDNDC3AVCNFSM6AAAAAA67R3DROVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMBRGM3DAOJQGA . You are receiving this because you commented.Message ID: @.***>