This folder contains files used for reference in the project — specifically, variant filtration and phenotype selection. Variant filtration is performed by a script in the scripts folder. Traits were selected by filtering on sample size from genome-wide association studies used for the Global Biobank Engine, and were then subject to manual curation. Both processes are described in detail in the manuscript.
- Variants included:
variant_qc_v2.prune.in - Traits included:
final_sumstats_v3.txt - Traits excluded:
blacklist.txt - Traits in final DeGAs instance:
final_phe_codes_v2.txt
The above files for traits often reference traits by their Global Biobank Engine IDs, which are not always human readable. For example, HC382 corresponds to Asthma, which was defined using both hospital records and verbal questionnaire data as part of a digital phenotyping project. The inclusion list also includes columns for phenotype name (string, column 2) and phenotype N (int, column 2, pooled across train, dev, and test sets in the UK Biobank white British cohort).
NB: Due to phenotype inclusion criteria, not all analysis traits present in final_sumstats_v3.txt were present in the final DeGAs instance (final_phe_codes_v2.txt), due to lack of genetic associations at p<1e-6 and to manual curation.