atgu / ukbb_pan_ancestry

Analyses conducting GWAS across the UKBB diverse superpopulations
MIT License
47 stars 23 forks source link

Updated sumstat exporting to use updated meta analysis results. Summa… #64

Closed rahulg603 closed 2 years ago

rahulg603 commented 2 years ago

…ry statistics are now outputted with *_hq columns if a "high quality ancestry-trait pair" meta-analysis was performed for that phenotype. Also added the option to exponentiate p-values on export, or leave them as natural logged. A new load_meta_analysis_results() function should now be used to load in meta-analyses.

rahulg603 commented 2 years ago

A couple of notes:

  1. For consistency, I have kept the original "pval_heterogeneity" column as is (even though all other meta analysis columns use the "_meta" suffix. The corresponding high quality version is "pval_heterogeneity_hq".
  2. I have tested this export on phenocode 30740 which you can look at here: gs://ukb-diverse-pops/ld_prune/211202_test/30740/quant/AFR-AMR-CSA-EAS-EUR-MID_batch2/biomarkers-30740-both_sexes-irnt.tsv.bgz
rahulg603 commented 2 years ago

Should have resolved all comments, and rewrote loo exporting function. See gs://ukb-diverse-pops/ld_prune/loo/sumstats/hq/211207_test/batch1 for some loo flat files I exported as a test (these are all 7 phenotypes with >= 5 ancestries passing hq filters).

rahulg603 commented 2 years ago

OH and also updated the lambda GC filter to allow lambda_gc < 5