search

cozygene / bisque

An R toolkit for estimation of cell composition from bulk expression data
68 stars 20 forks source link

Min number of markers in eset? #22

Open sjmonkley opened 3 years ago

sjmonkley commented 3 years ago

I have run the marker-based decomposition successfully with >1000 markers in the list with between 3 and 600 per cell type . I am now running the same eset with between 5-10 markers per cell type (85 in total) and I get the error:

Error in BisqueRNA::MarkerBasedDecomposition(bulk.eset = alvFC_CTR.norm.eset.ENSG, : No overlapping genes between markers and bulk.eset

This is despite running

length(intersect(ENSG.unique_slim$gene, featureNames(alvFC_CTR.norm.eset.ENSG)))#81

and getting 81 overlapping genes

So I am wondering is there an lower limit on number of overlapping genes?

brandonjew commented 3 years ago

Hi @sjmonkley, thanks for your interest in our method! The default minimum number of genes needed for each cell type is 5 (you can change this with the min.gene parameter in MarkerBasedDecomposition). Note that this value describes the minimum number of unique genes required for a cell type, which may be the source of your issue.

sjmonkley commented 3 years ago

Hi and thanks for the reply- I do not think that is the problem. from the overlapping 81 genes there are at least 5 per cell type as shown by counting number of genes per cluster below.

alvFC_CTR_intersect_count

A tibble: 11 x 2

cluster n

1 AberrantBasaloid 8 2 AT1 9 3 AT2 9 4 Basal 9 5 Cilitated 8 6 Differentiating 5 7 Goblet 5 8 Proliferating 10 9 SCGB3A2+ 8 10 SCGB3A2+ SCGB1A1+ 6 11 Transitional AT2 5`