search

broadinstitute / infercnv

Inferring CNV from Single-Cell RNA-Seq
Other
557 stars 164 forks source link

The expression quantity is inconsistent with the result of CNV #501

Open callaL opened 1 year ago

callaL commented 1 year ago

Hi @GeorgescuC,

First of all, thanks for this great tool!

I encountered a problem when using inferCNV.The expression of a gene in sample A is higher than the reference control(Fig1).However, when using inferCNV for detection, inferCNV detected a copy number deletion in the region where the gene is located.I checked the results of inferCNV and found that the trend of expression level was changed in the step of centralization(fig2,The input for drawing file is 11_recentered_cells_by_chrHMMi6.infercnv_obj)

fig1 a fig2 b

Do you have any suggestions to solve this problem? Thank you

GeorgescuC commented 1 year ago

Hi @callaL ,

As the name of the step implies, centering is done per cell during that step. The easiest way to illustrate its effects is to plot results after each step and compare plots after steps 10 and 11. With the example data this is what we get:

infercnv 10_smoothed_by_chr infercnv 11_centering_of_smoothed

You can see that some cells have overall lower or higher expression levels after step 10 due to varying sequencing depth, and, in the example, they tend to correlate by sample.

Regards, Christophe.