highlight cells in FeaturePlot with expression of two or more genes

[rahulnutron]

I am trying to identify cells which simultaneously express (if any) two or more genes of my choice. The expression may depend on a cut-off value. This can help me to identify a cell type. However, in feature plot I can see cells/cluster for only one gene. Is it possible to obtain such cell's information and marker genes only for these cells?

[mojaveazure]

Hi Rahul,

You can visualize the co-expression of two features by passing blend = TRUE; this will generate four plots: expression of the first feature, expression of the second, expression of the two together, and a heatmap key

FeaturePlot(object = pbmc_small, features = c('MS4A1', 'CD79B'), blend = TRUE)

[fly4all]

Hi. Following up on the same question, is there a way to see expression of more than 2 genes together?

[rahulnutron]

Thanks Paul! I need to update Seurat as I got an error
unused argument (blend = TRUE)

[rahulnutron]

This is useful. However, I want to do something different. I have 6 clusters. One of these clusters has 3-4 cells of rare cell-type. I can see that when I used FeaturePlot for three genes. All three genes has expression in some particular cells (however expression varies). I want to highlight the same cells for these three genes in the same plot (i.e., all cells having these three genes expressed simultaneously should be highlighted). The lower expression can be determined by some cut-off value. Then I want to perform DE analysis with these cells vs all others to find markers of this rare cell-type.

Thanks for help!

I have a code but its the same pipeline as given in the website. I just cant go beyond the FeaturePlot.

[mojaveazure]

Hi Maria and Rahul,

Currently, Seurat does not offer a way to visualize coexpression of more than two features. If you want to select cells based on expression of features, you can use WhichCells and pass the expression cutoffs the same way you would when subsetting a data frame (no quotes, backticks when calling features that have dashes in their names)

> # An example using the PBMC 3k object from our PBMC 3k tutorial
> WhichCells(object = pbmc_small, expression = MS4A1 > 3 & CD79A > 3 & CD3E > 3)
[1] "TTGGTACTGAATCC" "CATCATACGGAGCA" "TACATCACGCTAAC"

You could then take these cells, change the identity of them (see ?Idents for more details), and use that for differential expression.

[Yale73]

@mojaveazure Hi mojaveazure, I found we have the "overlay" parameter in FeaturePlot for Seurat v2.3.4, which can put two genes together in one featureplot, is it possible to recover this function, or do we upgrade this parameter to "blend"?

Thanks,
Yale