Skip to contents

Medusa: Mendelian Estimation in Distributed Standardized Analytics

Medusa-package.Rd

Medusa implements two-sample Mendelian Randomization (MR) within the OHDSI ecosystem using the OMOP Common Data Model as the data substrate. The package enables federated causal inference across distributed health networks without requiring individual-level data to leave any site.

The core methodological innovation is one-shot federated pooling via profile likelihood aggregation: each site computes a log-likelihood profile across a grid of parameter values and shares only that vector of numbers. The coordinator sums profiles across sites to obtain a pooled estimate without any iterative communication protocol.

Details

The analysis pipeline consists of nine modules:

  1. Instrument Assembly (getMRInstruments): Query OpenGWAS for GWAS summary statistics and apply LD clumping.

  2. Cohort Extraction (buildMRCohort): Extract outcome cohorts and genotype data from OMOP CDM sites.

  3. Covariate Assembly (buildMRCovariates): Assemble covariates via FeatureExtraction for adjustment and diagnostics.

  4. Instrument Diagnostics (runInstrumentDiagnostics): Validate instruments via F-statistics, PheWAS, allele-frequency checks, missingness summaries, and a placeholder negative-control interface.

  5. Outcome Model (fitOutcomeModel): Fit the binary outcome model and evaluate an exact or penalized profile log-likelihood on a grid.

  6. Likelihood Pooling (poolLikelihoodProfiles): Aggregate site-level log-likelihood profiles via pointwise summation.

  7. MR Estimation (computeMREstimate): Compute Wald ratio estimate with delta method standard errors.

  8. Sensitivity Analyses (runSensitivityAnalyses): IVW, MR-Egger, weighted median, Steiger filtering, leave-one-out.

  9. Reporting (generateMRReport): Generate self-contained HTML report with all results and diagnostics.

References

Davey Smith, G., & Hemani, G. (2014). Mendelian randomization: genetic anchors for causal inference in epidemiological studies. Human Molecular Genetics, 23(R1), R89-R98.

Bowden, J., Davey Smith, G., & Burgess, S. (2015). Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. International Journal of Epidemiology, 44(2), 512-525.

Bowden, J., Davey Smith, G., Haycock, P. C., & Burgess, S. (2016). Consistent estimation in Mendelian randomization with some invalid instruments using a weighted median estimator. Genetic Epidemiology, 40(4), 304-314.

Author

Maintainer: Adam Black black@ohdsi.org