With the expanding generation of large-scale biological datasets, there has been an ever-greater concern in understanding the reproducibility of discoveries and findings in a statistically reliable manner. We review several concepts in reproducibility and describe how one can adopt a multiple testing perspective on the problem. This leads to an intuitive procedure for assessing reproducibility. We demonstrate application of the methodology using RNA-sequencing data as well as metabolomics datasets. We will also outline some further problems in the field.
This is joint work with Daisy Philtron, Yafei Lyu and Qunhua Li (Penn State) and Tusharkanti Ghosh, Weiming Zhang and Katerina Kechris (University of Colorado).