Feature Request: Significance Without Class Labels

[DarioS]

Could there be a mode that doesn't require classes? I am trying to use the enrichment results to label cell types.

[ncborcherding]

Hey Dario,

What do you mean classes?

The enrichment calculations are done outside of classes - you can analyze the enrichment scores by any categorical variable (like clusters or cell type assignment).

Nick

[DarioS]

What I mean is that the result is a matrix of cells by gene set scores. Could a same dimension matrix be made for p-values? I think getSignificance is for two-group comparisons but I wonder if each single cell can have a p-value for each pathway.

[ncborcherding]

Hey Dario,

There are methods for annotation that will give you a confidence score based on a reference atlas (below are 2 good ones for R). Using enrichment scores might be an orthogonal method - but it is not asking which cell type is most likely or assigning a p-value, it is really saying across the ranked transcriptome, where are these gene lying.

1. Azimuth
2. Singler