Closed liu-xingliang closed 4 years ago
Running command:
infercnv_obj = CreateInfercnvObject(raw_counts_matrix=paste0('../../rerun.v140.meganormal/', patient, '.', tp, ".matrix"), annotations_file=paste0('../', patient, '.', tp, ".anno"), delim="\t", gene_order_file="../../../refdata-cellranger-GRCh38-3.0.0.gene_pos.chr_prefix.txt", ref_group_names = c('N_L', 'N_RB'))
infercnv_obj = infercnv::run(
infercnv_obj, cutoff=0.1, out_dir=paste0(patient, '.', tp), cluster_by_groups=T, denoise=T, resume_mode=F, no_prelim_plot = FALSE, HMM=TRUE, BayesMaxPNormal=0.5, diagnostics = TRUE, num_threads=8
)
Hi @liuxl18-hku ,
Could you try updating to the github version of the code by running:
library("devtools")
devtools::install_github("broadinstitute/infercnv", ref="RELEASE_3_10")
And then try to run the second command again?
Regards, Christophe.
@GeorgescuC ,
Thank you. After installing using your command, sessionInfo()
shows infercnv_1.2.2
, it's even lower than my previous version infercnv_1.4.0
which I got from Bioconductor.
I am re-running my commands using re-installed version.
bless~
Hi @liuxl18-hku
My bad, I copied the commands from the wiki but forgot to change the target branch. If you run the following, the version should show 1.4.0 again, but it has additional commits compared to the version on BioConductor.
library("devtools")
devtools::install_github("broadinstitute/infercnv", ref="master")
Regards, Christophe.
@GeorgescuC ,
Nevermind, I just updated my installation ending with infercnv_1.5.0. Now I am now testing it on a 19838 gene x 8315 cells dataset with four cores for running Bayesian Network Model on HMM predicted CNV's with 250Gb memory allocated.
It seems to take quite a while to run. I didn't use more cores as it seems to consume more memory and made my last few test running crashes.
Thank you.
Hi @liuxl18-hku ,
For 8315 cells and 19838 genes, 250GB of memory should be more than enough. Are you running on a cluster grid, or using a tool that could limit the memory allowed for a job (such as running in docker)?
Parallelization is enabled for the random trees subclustering if that option is enabled, and the Bayesian filtering only at this time.
Regards, Christophe.
@GeorgescuC
Yes, thank u, thd job done successfully after a few hrs.
Error messages:
Session info:
Full message: