Error for N: case-control study while using beta-SE option

[ganesh-chauhan]

Hi,

I have case-control GWAS summary statistics which does not have the N-columns. I tried using the "--use-beta-se" option, but I am getting an error that N was not computed. I guess I still need to provide the N column. (i) If I use max-N reported for the study, this will vary from SNP to SNP. (ii) should I use N-cases or N-total if I get the N-column some how. (iii) I tried calculating the effective sample size using the formula mentioned below, but approximating variance of phenotype to 1 does not give good result.

SE ~= (VP/(ES×2×p(1-p)))^1/2 [VP=variance of phenotype; ES=effective sample size; p=allele frequency; SE=standard error for beta].

The log for my analysis is presented below. Can you please suggest some way to proceed further. Thanks.

Calling ./mtag.py
--se-name StdErr
--bpos-name position
--a2-name Allele2
--n-min 0.0
--a1-name Allele1
--snp-name MarkerName
--use-beta-se
--eaf-name Freq1
--sumstats pheno1.out,pheno2.out,pheno3.out
--beta-name Effect
--out /home/ganesh/ganesh_data/pheno_3_test.mtag

2019/06/06/03:05:25 PM Beginning MTAG analysis... 2019/06/06/03:05:25 PM MTAG will use the provided BETA/SE columns for analyses. 2019/06/06/03:05:43 PM Read in Trait 1 summary statistics (8099789 SNPs) from pheno1.out ... 2019/06/06/03:05:43 PM <><><<>><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><> 2019/06/06/03:05:43 PM Munging Trait 1 <><><<>><><><><><><><><><><><><><><><><><><><><><><><><><>< 2019/06/06/03:05:43 PM <><><<>><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><> 2019/06/06/03:05:43 PM ERROR converting summary statistics:

2019/06/06/03:05:43 PM Traceback (most recent call last): File "/home/ganesh/MTAG/mtag/mtag_munge.py", line 817, in munge_sumstats raise ValueError('Could not determine N.') ValueError: Could not determine N.

[paturley]

I believe that when you use the beta/SE option, the N column doesn't get used in any of the statistical analyses, so you should be OK just putting in the max N or some other placeholder. Maybe confirm by running with two different values of N and verifying that the results are the same.

On Thu, Jun 6, 2019 at 7:57 AM ganesh-chauhan [email protected] wrote:

Hi,

I have case-control GWAS summary statistics which does not have the N-columns. I tried using the "--use-beta-se" option, but I am getting an error that N was not computed. I guess I still need to provide the N column. (i) If I use max-N reported for the study, this will vary from SNP to SNP. (ii) should I use N-cases or N-total if I get the N-column some how. (iii) I tried calculating the effective sample size using the formula mentioned below, but approximating variance of phenotype to 1 does not give good result.

SE ~= (VP/(ES×2×p(1-p)))^1/2 [VP=variance of phenotype; ES=effective sample size; p=allele frequency; SE=standard error for beta].

The log for my analysis is presented below. Can you please suggest some way to proceed further. Thanks.

Calling ./mtag.py --se-name StdErr --bpos-name position --a2-name Allele2 --n-min 0.0 --a1-name Allele1 --snp-name MarkerName --use-beta-se --eaf-name Freq1 --sumstats pheno1.out,pheno2.out,pheno3.out --beta-name Effect --out /home/ganesh/ganesh_data/pheno_3_test.mtag

2019/06/06/03:05:25 PM Beginning MTAG analysis... 2019/06/06/03:05:25 PM MTAG will use the provided BETA/SE columns for analyses. 2019/06/06/03:05:43 PM Read in Trait 1 summary statistics (8099789 SNPs) from pheno1.out ... 2019/06/06/03:05:43 PM <><><<>><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><> 2019/06/06/03:05:43 PM Munging Trait 1 <><><<>><><><><><><><><><><><><><><><><><><><><><><><><><>< 2019/06/06/03:05:43 PM <><><<>><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><> 2019/06/06/03:05:43 PM ERROR converting summary statistics:

2019/06/06/03:05:43 PM Traceback (most recent call last): File "/home/ganesh/MTAG/mtag/mtag_munge.py", line 817, in munge_sumstats raise ValueError('Could not determine N.') ValueError: Could not determine N.

— 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/72?email_source=notifications&email_token=AFBUB5KGNHU3UTLJDZXZNMTPZD3RLA5CNFSM4HVBHRL2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4GYACIFQ, or mute the thread https://github.com/notifications/unsubscribe-auth/AFBUB5NCRSN7RQ7ZL7UW7WTPZD3RLANCNFSM4HVBHRLQ .