search

broadinstitute / infercnv

Inferring CNV from Single-Cell RNA-Seq
Other
556 stars 163 forks source link

Cell based CNV state #175

Closed mea2712 closed 4 years ago

mea2712 commented 5 years ago

Hi! I'm using v1.1.2 and am after cell based CNV states. When I look at run.final.infercnv_obj@expr.data, I get a matrix of ~30000 rows (genes) and ~300 columns (cells): A10 A12 A13 A14 RP11-34P13.3 0.9183132 0.9829124 1.092044 0.9491200 FAM138A 0.9186002 0.9826138 1.091475 0.9508848 OR4F5 0.9186876 0.9821755 1.091720 0.9538763 RP11-34P13.7 0.9186683 0.9816884 1.091959 0.9579894 RP11-34P13.14 0.9189850 0.9816185 1.092707 0.9622377

What I would like to get is a matrix with cnvs as rownames and state assignments in the cells. Is this possible? Thank you!

GeorgescuC commented 4 years ago

Hi @mea2712 ,

If you run inferCNV with the HMM predictions, there are output tables that contain the predicted CNVs in a format similar to what you are looking for, though as they use clusters IDs, you will need to cross it with the infercnv_obj@tumor_subclusters information to get the cell IDs. The files to look for end in ".pred_cnv_genes.dat" or ".pred_cnv_regions.dat". For the unfiltered HMM predictions, the file names start with "17_HMM_predHMMi6", and for the Bayesian network-filtered results, they start with "HMM_CNV_predictions.HMMi6".

Regards, Christophe.