An R package for
Wang, L., Lin, Y., & Zhao, H. (2024). False Discovery Rate Control via Data Splitting for Testing-after-Clustering (arXiv:2410.06451). arXiv. https://doi.org/10.48550/arXiv.2410.06451
The proposed approach addresses the double-dipping issue in testing-after-clustering tasks, particularly in single-cell data analysis, where the same data is used both for clustering (to identify cell types) and for testing (to select differentially expressed genes), which can inflate false positives.
The xkcd-style cartoon is drawn with the help of R package xkcd
For single-cell data analysis, this package depends on Seurat. Since there are potential compatibility issue across different major versions of Seurat. We recommend Seurat v4, and it can be installed via the following command:
devtools::install_github("satijalab/seurat@v4.4.0")then you can install our package via
devtools::install_github("szcf-weiya/SplitClusterTest")For potential compatibility issues, you may need to install Matrix with version 1.6-5 or an older one.
- The Julia implementation: https://github.com/szcf-weiya/SplitClusterTest.jl
- For the comparison between data splitting and data fission, check https://github.com/szcf-weiya/fission_vs_splitting.
- The R code of the data splitting procedure in the regression setting: https://github.com/Jeremy690/DSfdr