getMarkerGenes discrepancy with FindAllMarkers in Seurat

[abedkurdi]

Hello guys,
I used your tool to visualize dynamically the clusters and it is awesome. However, when I called getMarkerGenes using the same exact parameters I used in FindAllMarkers in Seurat, It gave me way more genes. So for example, using a resolution of 0.4, Seurat gave me 555 genes in cluster 0, however, Cerebro gave me 666 genes in cluster 0. All genes found by FindMarkerGenes are present in the output of getMarkerGenes provided by your app.

So what do you think could be the difference between the two calls? I have checked the getMarkerGenes R script and it is based on FindAllMarkers from Seurat. Any idea what could be causing such discrepancy?

Thank you in advance.

[romanhaa]

Hi @abedkurdi!

Apologies for the late reply, but due to a lack of time I cannot maintain cerebroApp anymore. The issue you report is certainly strange, especially if you are sure that all parameters are identical between Seurat::FindAllMarkers() and cerebroApp::getMarkerGenes() (just to be sure, see this reference for more details). Since all genes of the shorter list are contained in the longer list, it sounds like it might have to do with filtering.

Cheers,
Roman