POP-Inf_analysis

This repository contains the analysis code for POP-Inf paper: Assumption-lean and data-adaptive post-prediction inference.

Simulations

Simulation scripts are included in the folder simulations. Results and figures can be reproduced via the following three steps:

1. Run simulations for mean estimation and OLS coefficient estimation in 1-1_sim_mean.Rmd and 1-2_sim_ols.Rmd, respectively. These scripts will automatically create a subfolder results containing raw results.
2. Summarize simulation results into csv files: 2-1_sum_mean.Rmd and 2-2_sum_ols.Rmd.
3. Reproduce simulation figures in the manuscript: 3_make_figure.Rmd.

Real data analysis

Our GTEx real data analysis is based on the codes and data used in Hypergraph factorization for multi-tissue gene expression imputation and The impact of sex on gene expression across human tissues. We thank the authors for sharing the codes.

There are five steps for the pipeline:

1. Clean the GTEx data from data downloaded from GTEx: 1_clean_gtex.R.
2. Run the HYFA to impute the gene expression: 2_run_hyfa.sh.
3. Process the observed and imputed gene expression: 3_process_gtex.R.
4. Run POP-Inf and compare it with alternative approaches to detect sex-biased genes with imputed gene expression: 4_popinf.R.
5. Make the Figure3 in the POP-Inf paper: 5_make_figure.R.

Contact

Please submit an issue or contact Jiacheng (jiacheng.miao@wisc.edu) or Xinran (xinran.miao@wisc.edu) for questions.

Reference

@article{miao2023assumption,
  title={Assumption-lean and Data-adaptive Post-Prediction Inference},
  author={Miao, Jiacheng and Miao, Xinran and Wu, Yixuan and Zhao, Jiwei and Lu, Qiongshi},
  journal={arXiv preprint arXiv:2311.14220},
  year={2023}
}