search

JonJala / mtag

Python command line tool for Multi-Trait Analysis of GWAS (MTAG)
GNU General Public License v3.0
169 stars 54 forks source link

Understanding FDR results #81

Open Lassana1 opened 4 years ago

Lassana1 commented 4 years ago

Hi, I ran MTAG on two traits and these are my FDR results:

2019/10/15/09:11:29 AM Beginning maxFDR calculations. Depending on the number of grid points specified, this might take some time... 2019/10/15/09:11:29 AM T=2 2019/10/15/09:11:29 AM Number of gridpoints to search: 207 2019/10/15/09:11:29 AM Performing grid search using 1 cores. 2019/10/15/09:11:29 AM Grid search: 10.0 percent finished for . Time: 0.001 min 2019/10/15/09:11:29 AM Grid search: 20.0 percent finished for . Time: 0.002 min 2019/10/15/09:11:29 AM Grid search: 30.0 percent finished for . Time: 0.002 min 2019/10/15/09:11:29 AM Grid search: 40.0 percent finished for . Time: 0.003 min 2019/10/15/09:11:30 AM Grid search: 50.0 percent finished for . Time: 0.004 min 2019/10/15/09:11:30 AM Grid search: 60.0 percent finished for . Time: 0.006 min 2019/10/15/09:11:30 AM Grid search: 70.0 percent finished for . Time: 0.008 min 2019/10/15/09:11:30 AM Grid search: 80.0 percent finished for . Time: 0.009 min 2019/10/15/09:11:30 AM Grid search: 90.0 percent finished for . Time: 0.010 min 2019/10/15/09:11:30 AM Grid search: 100.0 percent finished for . Time: 0.011 min 2019/10/15/09:11:30 AM Saved calculations of fdr over grid points in mtag_results/dsst.sevltrecall/p1p4/p1p4_fdr_mat.txt 2019/10/15/09:11:30 AM <><><<>><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><> 2019/10/15/09:11:30 AM grid point indices for max FDR for each trait: [18 40] 2019/10/15/09:11:30 AM Maximum FDR 2019/10/15/09:11:30 AM Max FDR of Trait 1: 5.07817107562e-06 at probs = [ 0. 0.1 0.8 0.1] 2019/10/15/09:11:30 AM Max FDR of Trait 2: 0.0128497124847 at probs = [ 0. 0.5 0.3 0.2] 2019/10/15/09:11:30 AM <><><<>><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><> 2019/10/15/09:11:30 AM Completed FDR calculations.

1) Can you please explain how to interpret these results? 2) How do I run maxFDR on each of the sets of summary statistics separately? (As suggested in #61 )

Thanks in advance.

Lassana1 commented 4 years ago

To be more specific, I understood FDR as a fixed rate of discovery for a trait. So what does the probs=[0. 0.1 0.8 0.1] mean?

Lassana1 commented 4 years ago

Hi :) any help is appreciated here!

paturley commented 4 years ago

Ah. Apologies. It's been a bit hectic since I got back.

maxFDR corresponds to the maximum that FDR could be if the beta coefficients are drawn from a spike-and-slab distribution. This is described in a little detail in the methods section of the MTAG paper and in greater detail in the online materials. Since the actual distribution of effect sizes is unknown, it's meant to be a worst-case scenario under some assumptions.

The probabilities are the a weights associated with the fraction of SNPs for which both SNPs are associated, neither are associated, or just one is, though I don't recall the order off the top of my head. I think you can look it up in the documentation.

On Thu, Oct 24, 2019 at 3:13 PM Lassana1 notifications@github.com wrote:

Hi :) any help is appreciated here!

— 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/81?email_source=notifications&email_token=AFBUB5OX5YGCDQY4KPYEQFLQQHXVTA5CNFSM4JA6CQWKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOECGD4GQ#issuecomment-546061850, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFBUB5OMZWWHVLHMVVADGNDQQHXVTANCNFSM4JA6CQWA .