search

broadinstitute / infercnv

Inferring CNV from Single-Cell RNA-Seq
Other
557 stars 164 forks source link

A sudden pause at STEP 7 #484

Closed judyyyyhsgfh closed 1 year ago

judyyyyhsgfh commented 1 year ago

Hello, thank you for developing and maintaining this wonderful tool.

I tried to get the output file "17_HMM_predHMMi6.hmm_mode-samples.cell_groupings" containing CNV classifications, but the process paused at STEP 7 without warnings or errors.

Here are my codes : ‘ infercnv_obj = CreateInfercnvObject(raw_counts_matrix=exprMatrix, annotations_file=" groupFiles.txt", delim="\t", gene_order_file= "geneLocate.txt", ref_group_names="0")

infercnv_obj = infercnv::run(infercnv_obj, cutoff=1, out_dir="output1202", cluster_by_groups=TRUE, tumor_subcluster_partition_method='random_trees', tumor_subcluster_pval=0.05, analysis_mode="subclusters", denoise=TRUE, HMM=TRUE, num_threads=4) ’

It paused here as the picture shows, and the occupations of CPU and storage went down.

image

Version informations: R version 4.2.1 (2022-06-23 ucrt) Platform: x86_64-w64-mingw32/x64 (64-bit) Running under: Windows 10 x64 (build 22000) infercnv: 1.12.0

Thank you so much for your help! Judy

GeorgescuC commented 1 year ago

Hi @judyyyyhsgfh ,

The random trees subclustering method, which is the step you are at, can take a while to run, especially if your dataset has over a couple thousand cells. If it takes too long to run, you can try the default subclustering method, Leiden, which is a lot faster and has more fine tuning options available.

Regards, Christophe.

judyyyyhsgfh commented 1 year ago

Hi @GeorgescuC , thank you so much for your reply!

I tried the default analysis mode at first (without codes 'analysis_mode="subclusters" '). It seems like that there's no branching information about subclusters in the output file "17_HMM_predHMMi6.hmm_mode-samples.cell_groupings", and that's what I tried to get. I ran codes above on another computer with larger RAM (16GB to 128GB), and got all output files in about 10 hours.