Development roadmap

Support more studies Partial

Current support is generic at the dataframe level. Numeric outcomes, predictors, and covariates can come from any study.

Next: logistic regression, survival models, categorical handling, and adapters for common cohort exports.

Faster PheWAS execution Planned

The current implementation performs one statsmodels fit per predictor-outcome pair.

Next: residualize covariates once, batch regressions with NumPy, benchmark memory and runtime, then add parallel execution only where it improves measured workloads.

More visualizations Partial

Manhattan, volcano, and category enrichment plots are implemented with deterministic options and result tables.

Next: QQ plots, forest plots, effect-comparison plots, and interactive output when a concrete analysis needs them.

BioThings and nearby genes Partial

BioThings clients provide functional variant and gene summaries. Nearby-gene mapping uses caller-supplied genomic intervals.

Next: add an assembly-aware coordinate service for nearby genes while keeping local interval files as an offline option.

More MR methods Partial

Eleven estimators are implemented and their numerical behavior has been compared with TwoSampleMR and MR-PRESSO.

Next: prioritize methods with primary references, clear assumptions, maintained comparison implementations, and real user demand.

Robust harmonization Improved

Current harmonization validates one canonical schema, normalizes chromosome and allele values, aligns allele pairs and strand complements, flips reversed effects, resolves duplicates, and removes incompatible or ambiguous rows.

Next: allele-frequency handling for palindromic variants, normalized indel representation, multiallelic validation, and explicit harmonization reports.

Documentation Improved

The site now covers workflows, schemas, API usage, MR method selection, plotting examples, and paper reproduction.

Next: add focused tutorials and generated signatures when the public API changes.