Statistical Tools for Replicable and Reproducible Research

Statistical Tools for Replicable and Reproducible Research Recent technology advances have changed the practice of scientific research. More studies are conducted in each research field, and large-scale data are recorded in each study, such that many hypotheses are examined. Data processing and analysis pipelines become more complex, and there are convenient tools for storing and distributing complete databases and code libraries. Three related aspects of the scientific research have been influenced by these changes and gained attention in the scientific community: The control over the rate of false discoveries in large-scale studies; the ability to replicate findings from previous studies and strengthen their conclusions; and the ability reproduce analysis results, given the original data and code. Three applications of replicable and reproducible research are presented in this work: An R package for replicability analysis in the empirical Bayes framework; a community-based system for replicability assessment across laboratories, including a web application; and reproducible analysis pipeline for rare events differences testing.