Mapping ICD10 codes from field 41270 via PheWAS R package using createPhenotypes results in only ~1250 phenotypes. Does anyone know how to get ALL the phenotypes in the UKB - but particularly all related to mental disorders?
Hi UKB community,
Has anyone managed to run the PheWAS package and get all the phenotypes in the UKB? I've been following the DNAnexus end_to_end_gwas_phewas tutorial, I successfully create the phenotypes file, I have my covariates file and the genotype (I'm after one particular SNV) but when I map phenotypes to ICD10 codes using createPhenotypes command I only get ~1250 phenotypes:
# Map ICD10 codes to code description
phenotypes = createPhenotypes(
pheno, min.code.count=1, add.phecode.exclusions=T, translate=T, vocabulary.map=PheWAS::phecode_map_icd10
)
This is reflected later in a very "thin" Manhattan Plot. When creating the phenotypes table following get-phewas-data.ipynb, at the stage when pheno_long_all table is created, there are > 12000 unique phenotypes.
For reference, when running example code included in the package documentation there are ~1800 phenotypes, which is better but still not great. I would particularly like to map all the phenotypes related to mental/psychiatric conditions.
Could anyone potentially help?
Comments
4 comments
@Anastazie Sedlakova? is on vacation right now. I think she will be able to help here once she is back. Feel free to share with us any new findings you might have in the meantime.
Thanks @Ondrej Klempir? , will wait for @Anastazie Sedlakova? ?
{@005t0000008C5AmAAK}? I looked into PheWAS package and I think that problem is in createPhenotypes. When using phecode_map_icd10, it maps ICD10 to ICD9. ICD10 has about 69000, whereas ICD9 has only 14000 diseases. When I looked at this table, I see that a lot of different ICD10 codes have the same corresponding ICD9 code. In that table there were only 1571 disctinct ICD9 codes. I think that you can look and decide whether ICD9 granularity is OK for you. If not, I recommend trying another package for PheWAS. I know about PHESANT tool, however, I do not have experience with it.
ok thank you @Anastazie Sedlakova?
Please sign in to leave a comment.