Summarizing the outputs of inferCNV

[enBiter]

Dear authors,

I'm trying to make a summary of my inferCNV outputs, but stuck at solving these questions, any help would be appreciated. I'm using inferCNV version 1.17.0

1. What does the state column means in the HMM_CNV_predictions.HMMi6.rand_trees.hmm_mode-subclusters.Pnorm_0.5.pred_cnv_genes.dat tabular output? I see many genes with state 2, does that mean that the predicted CNV is diploid (unchanged)?

2. Is there a way to obtain the outputs in the form of GRanges? I'd like to plot a summary of CNV for subclones, similar to what Durante et al. show in Fig. 2b in the linked paper. I'd also like to make a graphical comparison of inferCNV output and CNVkit output from WES seq data of the same samples. I guess that GRanges is the best format for this task.

3. When running my analysis without setting the k_obs_groups parameter, I always get 8 subclones regardless of genomic structure of the sample. But when setting this parameter to a lower value, the number of cell groups in HMM_CNV_predictions.HMMi6.rand_trees.hmm_mode-subclusters.Pnorm_0.5.pred_cnv_genes.dat is still 8. How do I extract the genomic data for my set number of subclones?

My run parameters are as follows, feel free to point out anything that doesn't make sense:

infercnv::run(
    inferCNVobj,
    cutoff = 0.1,
    out_dir = inferCNV.sample.path,
    resume_mode = TRUE,

    cluster_by_groups = FALSE,
    cluster_references=FALSE,
    denoise = TRUE,

    k_obs_groups = k_obs_groups.list[[current_sample]],

    HMM = TRUE,
    HMM_type = 'i6',
    analysis_mode = "subclusters",
    hclust_method='ward.D2',

    tumor_subcluster_partition_method = "random_trees",
    tumor_subcluster_pval = 0.01,

    write_phylo = TRUE,

    no_prelim_plot = TRUE,
    output_format = "png",
    png_res = 600,

    num_threads = 16
  )

Please let me know if you need any other details.

Best regards

Dmitry