You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I encountered an issue while running the "run_numbat" function on a dataset of around 40,000 cells. The process was time-consuming, and unfortunately, the job was terminated at the step "Using 4 CNVs to construct phylogeny" due to the time limit of the slurm system (3 days). I attempted to rerun the script, but it restarted from the beginning instead of resuming from the breakpoint. Is there any way to continue the execution from where it left off?
`
if(!dir.exists(paste0(numbat_res_base, selected_patient,'/run_numbat_out'))){
dir.create(paste0(numbat_res_base, selected_patient,'/run_numbat_out'))
}
out = run_numbat(
as(mat_merge,"CsparseMatrix"), # gene x cell integer UMI count matrix has to be a sparse matrix
ref_internal, # reference expression profile, a gene x cell type normalized expression level matrix, made above
df_allele_merge, # allele dataframe generated by pileup_and_phase script from the API
genome = "hg38", # to change if using a different genome
t = 1e-5,
ncores = 20,
ncores_nni=20,
plot = TRUE,
min_LLR=3,
out_dir = paste0(numbat_res_base, selected_patient,'/run_numbat_out') #where you want to save the output
)
`
The text was updated successfully, but these errors were encountered:
Thanks for the issue. 40k cells is a record for us .. are these all from the same patient? Unfortunately, there's currently no built-in mechanism for resuming Numbat runs from where it left off. The only workaround might be either running with more threads or splitting the cells by sample. Installing the most recent scistreer (v1.2.0) may also help since it speeds up the phylogeny part.
I had a very similar issue with a unintended shutdown after run_numbat running for >2 days (35k cells). I thus decided to modify the code to allow resuming at the latest saved stop point (output files saved). As an added benefit it also allows to add additional iterations if the previous amount was not sufficient to reach convergence (without having to start over).
It may not be the most elegant fix, but it works and my interrupted run has now finished. But as I have never really done pull requests and it may not be the "optimal" way to do it, I have just placed the forked repository here for now: https://github.com/Terkild/numbat
Hi,
I encountered an issue while running the "run_numbat" function on a dataset of around 40,000 cells. The process was time-consuming, and unfortunately, the job was terminated at the step "Using 4 CNVs to construct phylogeny" due to the time limit of the slurm system (3 days). I attempted to rerun the script, but it restarted from the beginning instead of resuming from the breakpoint. Is there any way to continue the execution from where it left off?
`
`
The text was updated successfully, but these errors were encountered: