Abstract
Here we present the combiroc R package, for signatures refinement in high throughput omics. Based on a ROC-driven marker selection, it can be used to find powerful smaller sub-signatures from scRNAseq experiments and to annotate cells using fewer markers. Trained on PBMC dataset, combiroc found NK marker combinations with high cell-discriminating power, in agreement with human protein atlas and that were validated both computationally and experimentally on independent datasets.
Competing Interest Statement
The authors have declared no competing interest.
Footnotes
The main differences of this second, revised version of the preprint manuscript are the following: 1) New training datasets were used. In the initial training section we used a more carefully annotated dataset with cell clusters originated by a multimodal approach. 2) More testing datasets: we used a total of three testing datasets, on which we showed the ability to discriminate NK cells from other cell types 3) We added an entirely new section in which we demonstrated that markers selected with combiroc for NK cells are consistent with both results annotated in the Human Protein Atlas and with results of cytofluorimetric experiments performed ad-hoc in our lab. 4) Whole code of the package was revised and some bugs were fixed, moreover some functions were changed and it's now possible to convert single cell Seurat dataset into combiroc format. Supplementary material and GitHubs code repositories were updated accordingly 5) An additional author was acknowedged due to the new experiments made (see point #3) and a different title waas draft for this revised version. 6) A revised, more concise, title was used for this version
https://ingmbioinfo.github.io/combiroc/articles/combiroc_vignette_1.html
https://ingmbioinfo.github.io/combiroc/articles/combiroc_vignette_2.html