Automates GWAS with PLINK2 genotype QC and REGENIE two-step association testing.
npx skills add https://github.com/ClawBio/ClawBio --skill gwas-pipeline
Please help me install this Agent Skill. Skill: gwas-pipeline Source: https://github.com/ClawBio/ClawBio/tree/main/skills/gwas-pipeline Command: npx skills add https://github.com/ClawBio/ClawBio --skill gwas-pipeline
Running a genome-wide association study normally requires manually orchestrating PLINK2 quality-control filters and REGENIE's two-step regression, writing hundreds of lines of bash, and tracking dozens of parameters. This Skill automates the entire workflow from genotype files to publication-ready Manhattan plots, QQ plots, and lead variant tables with a full reproducibility log.
Run the GWAS pipeline in demo mode on the bundled REGENIE example dataset to produce a full association report with Manhattan and QQ plots.
High-intent search queries and answers about installing and using this skill.
Provide a PLINK BED prefix, a phenotype file, and optionally a covariate file, then run the pipeline with your trait name and type (bt for binary, qt for quantitative). It executes PLINK2 QC, REGENIE Step 1 and Step 2, and writes a report with plots and summary statistics.
The pipeline accepts PLINK binary files (.bed/.bim/.fam) for genotypes, BGEN v1.2+ files for Step 2 imputed variants, and plain-text phenotype and covariate files with FID and IID columns. A demo mode uses REGENIE's bundled 500-sample example dataset.
Yes. Binary traits use REGENIE's Firth logistic regression with the fast approximation and a p-value threshold of 0.01, while quantitative traits use standard linear regression. Select the mode with the --trait-type bt or qt flag.
Default thresholds follow Anderson et al. (2010): variant missingness 0.02, sample missingness 0.02, minor allele frequency 0.01, and HWE p-value 1e-6, plus LD pruning (window 1000, step 100, r2 0.9). All thresholds are configurable via command-line flags.
Both external binaries must be installed and on PATH, or present in a clawbio-gwas conda environment. Install them with conda create -n clawbio-gwas -c conda-forge -c bioconda plink2 regenie, then rerun the pipeline.
Outputs include a Markdown report, result.json with lambda GC and top variants, full summary statistics and lead variant TSV tables, Manhattan and QQ plots, and a reproducibility folder containing the executed commands and software versions.
jaechang-hitsRun PLINK2 GWAS association testing with quality control and principal component analysis.
Imad-OuteRetrieve SNP-trait associations and summary statistics from the NHGRI-EBI GWAS Catalog.
mims-harvardCompare GWAS studies, meta-analyze loci, and assess replication across cohorts.
mims-harvardQueries cross-ancestry PheWAS associations for a variant or gene across four biobanks.
