sc-fgsea performs cluster annotation of single-cell RNA sequencing data. The tool runs on the fGSEA framework and leverages existing cell-type specific genesets to enable itterative, supervised annotation based on computed p-values and enrichment scores.
The documented code for sc-fgsea is provided as an R markdown file, which can easily be loaded into RStudio.
This tool was developed by Alexandre Trapp in the context of Emmrich et al (2022):
Characterization of naked mole-rat hematopoiesis reveals unique stem and progenitor cell patterns and neotenic traits
Stephan Emmrich, Alexandre Trapp, Frances Tolibzoda Zakusilo, Maggie E Straight, Albert K Ying, Alexander Tyshkovskiy,
Marco Mariotti, Spencer Gray, Zhihui Zhang, Michael G Drage, Masaki Takasugi, Jan-Henning Klusmann, Vadim N Gladyshev,
Andrei Seluanov, Vera Gorbunova
Required data files (located in the data folder) are also provided to run the tool and reproduce the analysis.
The input to sc-fgsea consists of two components:
- a .txt markers file, outputted by Seurat's FindAllMarkers function
- a .gmt atlas file (see here for a description of this file format), containing a variety of genesets specific to particular cell types
Example markers and atlas files are provided in data:
- atlasV14.gmt
- hglBM.NOmp.markers.clust.txt
The output of sc-fgsea is an Excel file, containing p-values and enrichment scores for each geneset/cluster pair from the input marker file and atlas. An example output file (hglBM.cluster.markers.output.xlsx) can be found in the output folder, with the following format:
| pathway | pval | padj | ES | NES | nMoreExtreme | size | leadingEdge | cluster |
|---|---|---|---|---|---|---|---|---|
| ROSSI_mLTHSC | 0.009482058 | 0.039587592 | -0.854460094 | -1.738297467 | 384 | 3 | CALML4, UPP1, CLU | 1 |