infercnv error at STEP 18 - file descriptor is too large for select()

Question

[yura.grabovska]

I'm trying to run infercnv on a matrix of 10x data. In this example i'm using a dummy dataset with 500 'tumour' cells vs a reference of normal brain cells

This is the function call:

infercnv.obj <- infercnv::run(infercnv.obj,
                              cutoff = 0.1, 
                              out_dir = paste0("infercnv/", [e-mail]),
                              cluster_by_groups = TRUE,
                              denoise = TRUE,
                              hclust_method = "ward.D2",
                              scale_data = TRUE, 
                              HMM = TRUE, 
                              HMM_type = "i6",
                              analysis_mode = "subclusters")

It runs fine until it gets to this point:

    STEP 18: Run Bayesian Network Model on HMM predicted CNVs

INFO [2025-10-06 14:13:33] Creating the following Directory:  infercnv/CHCO061/BayesNetOutput.HMMi6.leiden.hmm_mode-subclusters
INFO [2025-10-06 14:13:33] Initializing new MCM InferCNV Object.
INFO [2025-10-06 14:13:33] validating infercnv_obj
INFO [2025-10-06 14:13:33] Total CNV's:  449
INFO [2025-10-06 14:13:33] Loading BUGS Model.
INFO [2025-10-06 14:13:33] Running Sampling Using Parallel with  4 Cores
Error in selectChildren(ac[!fin], -1) : 
  file descriptor is too large for select()

The support for infercnv on Github is on hiatus as of 2024 so I was wondering if anyone here has had this error and has overcome it?

Answer 1

[yura.grabovska]

Noticed that running infercnv::CreateInfercnvObject() generates a warning:

WARN [2025-10-06 14:43:38] Please use "options(scipen = 100)" before running infercnv if you are using the analysis_mode="subclusters" option or you may encounter an error while the hclust is being generated.

Setting that as advised seems to fix the issue