Plotting infercnv results

Question

I'm working with matched single cell data, where we have treated and untreated samples for the same patient. I ran CNV analysis using the infercnv package.

I've followed the tutorial:

# data matrix
counts_matrix <- scData@assays$RNA@counts
meta = data.frame(labels = Idents(scData), row.names = names(Idents(scData)))
unique(meta$labels) # check the cell labels

# Create the infercnv object
infercnv_obj <- CreateInfercnvObject(raw_counts_matrix = counts_matrix,
                                     annotations_file = meta,
                                     delim = "\t",
                                     gene_order_file = "hg38_gencode_v27.txt",
                                     ref_group_names = NULL)

# perform infercnv operations to reveal cnv signal
infercnv_obj = infercnv::run(infercnv_obj,
                             cutoff=1,  # use 1 for smart-seq, 0.1 for 10x-genomics
                             out_dir="output_dir",  # dir is auto-created for storing outputs
                             cluster_by_groups=T,   # cluster
                             denoise=T,
                             HMM=T
                             )

Now, I'm not sure how to plot the results. Any help would be appreciated.

When running the below recommendation I get this:

INFO [2023-05-19 12:49:22] ::plot_cnv:Start
INFO [2023-05-19 12:49:22] ::plot_cnv:Current data dimensions (r,c)=34108,17624 Total=82463956 Min=0 Max=55099.
INFO [2023-05-19 12:49:22] ::plot_cnv:Depending on the size of the matrix this may take a moment.
<sparse>[ <logic> ]: .M.sub.i.logical() maybe inefficient
Warning: sparse->dense coercion: allocating vector of size 4.5 GiBINFO [2023-05-19 12:50:12] plot_cnv(): auto thresholding at: (-2.725632 , 3.000000)
INFO [2023-05-19 12:50:35] plot_cnv_observation:Start
INFO [2023-05-19 12:50:35] Observation data size: Cells= 17624 Genes= 34108
INFO [2023-05-19 12:50:35] clustering observations via method: ward.D
INFO [2023-05-19 12:50:35] Number of cells in group(1) is 331
INFO [2023-05-19 12:50:36] group size being clustered:  331,34108
Error in parallelDist(data_to_cluster, threads = infercnv.env$GLOBAL_NUM_THREADS) : 
  x must be a matrix or a list of matrices.

Answer 1

If you haven't already, I recommend checking out the infercnv YouTube tutorial (https://github.com/broadinstitute/inferCNV/wiki/Running-InferCNV) for detailed instructions on running InferCNV. Additionally, you can explore Uphyloplot2 (https://github.com/harbourlab/uphyloplot2) for building a tree from the output generated by InferCNV.

I noticed that you didn't use the untreated sample as the normal reference for your CNV analysis. I'm curious to understand your reasoning behind this. In my case, I also have an untreated sample that isn't diploid, along with two treated samples. Initially, I planned to use the untreated sample as the "normal" reference. However, considering that InferCNV assumes a normal diploid reference, I'm now uncertain if using the untreated control would be appropriate.

Please let me know your thoughts, any help would be much appreciated!

Answer 2

Personally I would not use intercnv_obj throughout, because is there was a bug in line 2, i.e. using = instead of <- then the created intercnv_obj might be trying to be passed into plot_cnv.

Could you try

data(infercnv_obj) 

plot_cnv(
  infercnv_obj,
  out_dir = ".",
  title = "inferCNV",
  obs_title = "Observations (Cells)",
  ref_title = "References (Cells)",
  cluster_by_groups = TRUE,
  cluster_references = TRUE,
  plot_chr_scale = FALSE,
  chr_lengths = NULL,
  k_obs_groups = 1,
  contig_cex = 1,
  x.center = mean([email protected]),
  x.range = "auto",
  hclust_method = "ward.D",
  custom_color_pal = NULL,
  color_safe_pal = FALSE,
  output_filename = "infercnv",
  output_format = "png",
  png_res = 300,
  dynamic_resize = 0,
  ref_contig = NULL,
  write_expr_matrix = FALSE,
  write_phylo = FALSE,
  useRaster = TRUE
)

---

If the above doesn't work ...

What I think is happening is the input data is the problem. This precise error is already known and was reported on Broad's GitHub here:

https://github.com/broadinstitute/infercnv/issues/415

The responder (Broad Institute) believed it was the version problem, i.e. the version was outdated. However a separate respondee believed it was the input data format and suggested shifting away from a sparse matrix. That would make sense in context to the error being thrown.