Closed AlfredoKCL closed 5 years ago
For the cohorts in an analysis, their effect column should be all beta for quantitative traits, or all OR for case-control design.
In your case, one cohort uses "beta" and the other uses "or", and it is not actually allowed.
A possible solution is to transform "or" value to beta by taking the natural logarithm of the "or" values.
I added worked exampled at the end of the wiki page, https://github.com/gc5k/GEAR/wiki/LambdaMeta.
Kind regards, g
Dear developers,
I am trying to run "java -jar gear.jar lam --meta-batch sum_stat_files.txt --qt-size sum_stat_size.txt --out test" on two summary stats files to see if there are overlapping samples.
The header of the first file is: CHR SNP BP A1 A2 OR SE P And the header of the second is: CHR SNP BP A1 A2 BETA SE P
When I run the command I get:
[INFO] [INFO] The keyword for 'SNP' is set to snp [INFO] The keyword for 'BETA' is set to beta [INFO] The keyword for 'SE' is set to se [INFO] The keyword for 'A1' (reference allele) is set to a1 [INFO] The keyword for 'A2' (the other allele) is set to a2 [INFO] The keyword for 'CHR' is set to chr
[INFO] 2 cohorts are remained for analysis.
[INFO] Cannot find the beta/or column in SCZvsCONT_new.sumstats [INFO] GEAR quit.
Can you please help me with that?