Raw count or Normalized count?

[parkjooyoung99]

Dear developer,

I am trying to run infercnv with my data, and faced Error below at step 15.

Just in case I have run NormalizeData for my seurat object and re-create infercnv object but I don't get why I should do this, since i am using the "count" slot not the "data" slot for creating infercnv object.

Moreover, if I see the example data you have provided here , the counts do not seem like a raw count. I am confused with it, and hope to hear about it.

Thank you!

[parkjooyoung99]

Still having the same error and if I run Normalizedata, the below error pops out .

My code is infercnv_obj <- infercnv::run(infercnv_obj, cluster_by_groups = T, # cluster or not analysis_mode = "subclusters", out_dir = 'HMM_F', denoise = T, HMM = F, save_rds = T, plot_steps=F, save_final_rds = T, no_plot = F, useRaster = F )

Also, it seems some steps do not generate temporary rds files,

I wonder if this is abnormal.

You can find my object here

[GeorgescuC]

Hi @parkjooyoung99 ,

This error is due to a change in how Seurat works, but a new commit currently only available on Github should fix this issue, so you can update using the repo as source. Infercnv takes raw counts as input, and you should not provide already normalized data to it. This particular error message comes from converting infercnv's residual expression data to a Seurat object to be used for Seurat's nearest neighbor search for the Leiden clustering.

Regards, Christophe.