search

JonJala / mtag

Python command line tool for Multi-Trait Analysis of GWAS (MTAG)
GNU General Public License v3.0
173 stars 55 forks source link

Two questions #88

Open kys21207 opened 4 years ago

kys21207 commented 4 years ago
  1. I have case-control GWAS summary statistics that do not have the N-columns. How to define 'N'? Should I use N-cases or N-total to define 'N' ?
  2. Can I run MTAG chromosome by chromosome because of memory error? Any advice? Thank you
paturley commented 4 years ago
  1. I recommend that you use the effective N formula from https://www.nature.com/articles/nprot.2014.071.
  2. Running MTAG by chromosome is fine. The only consideration is that the Sigma and Omega matrix will vary by chromosome. Not a big deal, but I often recommend to users that they report the Sigma and Omega matrix in the supplements of their work, so you would just be reporting 22 such matrices.

On Wed, Mar 4, 2020 at 2:09 PM Kijoung Song notifications@github.com wrote:

  1. I have case-control GWAS summary statistics that do not have the N-columns. How to define 'N'? Should I use N-cases or N-total to define 'N' ?
  2. Can I run MTAG chromosome by chromosome because of memory error? Any advice? Thank you

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/omeed-maghzian/mtag/issues/88?email_source=notifications&email_token=AFBUB5I2NKIMG2YQVBB7NHTRF2RPNA5CNFSM4LBQ76IKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4ISQHR4Q, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFBUB5MO6ZEC5U2PBPRPEY3RF2RPNANCNFSM4LBQ76IA .

kys21207 commented 4 years ago

Thank you so much for your suggestion.

kys21207 commented 3 years ago

Hi Paturley,

I recently compared two MTAG results (one based on the whole genome and the other based on chromosomes). In general, the two Manhattan plots are similar. However, some chromosomes showed significant differences. For example, the signal of MTAG by chromosome showed higher power on several specific chromosomes and much lower power on one or two specific chromosomes. I understand this is due to the difference in the sigma and omega matrix of the chromosome. In this case, which MTAG result should I take? And I am still wondering if this is ok to run MTAG by chromosome?

paturley commented 3 years ago

Hi Kijoung,

Yes. The reason the results may differ is that Omega and Sigma are forced to be the same if you run it all together and they are allowed to vary across chromosomes if you run it separately by chromosomes. It's a bit difficult to say without seeing the data, but my gut instinct is that the results that are not split by chromosome will be more reliable since the estimates of Omega and Sigma will be estimated using more SNPs. It also is probably easier to describe what you are doing in an eventual manuscript since you can just report the Omega and Sigma that you are using one time rather than reporting different matrices for each chromosome.

Best, Patrick

On Thu, Dec 3, 2020 at 8:47 AM Kijoung Song notifications@github.com wrote:

Hi Paturley,

I recently compared two MTAG results (one based on the whole genome and the other based on chromosomes). In general, the two Manhattan plots are similar. However, some chromosomes showed significant differences. For example, the signal of MTAG by chromosome showed higher power on several specific chromosomes and much lower power on one or two specific chromosomes. I understand this is due to the difference in the sigma and omega matrix of the chromosome. In this case, which MTAG result should I take? And I am still wondering if this is ok to run MTAG by chromosome?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/JonJala/mtag/issues/88#issuecomment-738003868, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFBUB5IZDCPOAP6EHMKOBW3SS6JGDANCNFSM4LBQ76IA .