phewas

Example commands to run the pipeline:

nextflow run main.nf --agg_vcf_file_list testdata/vcf_samples.csv \
                     --input_phenofile testdata/cb_binary_pheno.phe \
                     --input_id_code_count testdata/phewas_id_code_count.csv \
                     --pheno_codes "icd10"

Alternatively, it is possible to specify plink input files:

nextflow run main.nf --input_phenofile "s3://lifebit-featured-datasets/pipelines/phewas/cb_binary_pheno.phe" \
                     --input_id_code_count "s3://lifebit-featured-datasets/pipelines/phewas/phewas_id_code_count.csv" \
                     --plink_input "s3://lifebit-featured-datasets/pipelines/biobank-gwas/testdata/sampleA.{bed,bim,fam}" \
                     --pheno_codes "icd10"

A covariate file can be used for pheWAS analysis (optional, but strongly recommended). Please note that the format for the covariate file should be a comma-delimited file with sample IDs in first column (id as first column name). The regression model used in pheWAS will use all supplied covariates by default :

id,sex,age...
1,1,25

nextflow run main.nf --input_phenofile "s3://lifebit-featured-datasets/pipelines/phewas/cb_binary_pheno.phe" \
                     --input_id_code_count "s3://lifebit-featured-datasets/pipelines/phewas/phewas_id_code_count.csv" \
                     --plink_input "s3://lifebit-featured-datasets/pipelines/biobank-gwas/testdata/sampleA.{bed,bim,fam}" \
                     --covariate_file "s3://eu-west-2-example-data/pipelines/phewas/testdata/cov.csv"
                     --pheno_codes "icd10"

--firth_regression parameter can supplied to run the association tests with Firth logistic regression - this is highly recommended for phenotypes with significantly imbalanced case/control ratios.

With colocalization analysis with binary GWAS:

nextflow run main.nf --input_phenofile "s3://lifebit-featured-datasets/pipelines/phewas/cb_binary_pheno.phe" \
                     --input_id_code_count "s3://lifebit-featured-datasets/pipelines/phewas/phewas_id_code_count.csv" \
                     --plink_input "s3://lifebit-featured-datasets/pipelines/biobank-gwas/testdata/sampleA.{bed,bim,fam}" \
                     --pheno_codes "icd10" \
                     --post_analysis "coloc" \
                     --gwas_input "s3://lifebit-featured-datasets/pipelines/biobank-gwas/gwas_summary_bin.csv" \
                     --gwas_trait_type "binary"

With colocalization analysis with quantitative GWAS:

nextflow run main.nf --input_phenofile "s3://lifebit-featured-datasets/pipelines/phewas/cb_binary_pheno.phe" \
                     --input_id_code_count "s3://lifebit-featured-datasets/pipelines/phewas/phewas_id_code_count.csv" \
                     --plink_input "s3://lifebit-featured-datasets/pipelines/biobank-gwas/testdata/sampleA.{bed,bim,fam}" \
                     --pheno_codes "icd10" \
                     --post_analysis "coloc" \
                     --gwas_input "s3://lifebit-featured-datasets/pipelines/biobank-gwas/gwas_summary_qt.csv" \
                     --gwas_trait_type "quantitative"

jpcartailler/PheWAS

phewas