Speeding up interval estimation for R2-based mediation effect of high-dimensional mediators via cross-fitting

Area of Science:

• Genomics
• Biostatistics
• Systems Biology

Background:

• Mediation analysis investigates how molecular phenotypes, like gene expression, link exposures to health outcomes.
• Traditional mean-based mediation measures can be inaccurate with high-dimensional omics data due to effect cancellation.
• Nonparametric bootstrap methods for R-squared mediation measures are computationally intensive.

Purpose of the Study:

• To develop a computationally efficient two-stage, cross-fitted estimation procedure for the R-squared total mediation effect measure.
• To provide a more reliable mediation analysis for high-dimensional molecular data.
• To enable accurate confidence interval estimation for R-squared mediation effects.

Main Methods:

• A two-stage, cross-fitted estimation procedure was developed for the R-squared mediation measure.
• Iterative Sure Independence Screening (iSIS) was used in subsamples to identify relevant mediators.
• Ordinary least squares regressions and a closed-form asymptotic distribution were used for variance and confidence interval estimation.

Main Results:

• The proposed procedure significantly improves computational efficiency compared to resampling methods.
• The method maintains comparable coverage probability to bootstrap methods.
• Application to the Framingham Heart Study validated findings on gene expression mediating blood pressure and identified gene roles in cholesterol levels.

Conclusions:

• The new estimation procedure offers a computationally efficient and accurate approach for R-squared mediation analysis in high-dimensional omics settings.
• This method enhances the understanding of molecular mechanisms underlying health outcomes.
• The procedure is available in the R package CFR2M.