The program got stuck when simulating spatial data

[chthub]

Thanks for your awesome tool. When I try the tutorial to simulate the spatial transcriptomic data, the code got stuck in "Start Parameter Extraction", here is the code:

library(scDesign3)
library(SingleCellExperiment)
library(ggplot2)
library(dplyr)
library(viridis)
theme_set(theme_bw())

example_sce <- readRDS((url("https://figshare.com/ndownloader/files/40582019")))
print(example_sce)

example_sce <- example_sce[1:10, ]

set.seed(123)
example_simu <- scdesign3(
  sce = example_sce,
  assay_use = "counts",
  celltype = "cell_type",
  pseudotime = NULL,
  spatial = c("spatial1", "spatial2"),
  other_covariates = NULL,
  mu_formula = "s(spatial1, spatial2, bs = 'gp', k= 10)",
  sigma_formula = "1",
  family_use = "nb",
  n_cores = 128,
  usebam = TRUE,
  corr_formula = "1",
  copula = "gaussian",
  DT = TRUE,
  pseudo_obs = FALSE,
  return_model = FALSE,
  nonzerovar = FALSE,
  parallelization = "pbmcapply"
)
print(example_simu)

Here is the outputs:

class: SingleCellExperiment
dim: 1000 2696
metadata(0):
assays(2): counts logcounts
rownames(1000): Calb2 Gng4 ... Fndc5 Gda
rowData names(0):
colnames(2696): AAACAAGTATCTCCCA-1 AAACACCAATAACTGC-1 ...
  TTGTTTCACATCCAGG-1 TTGTTTCCATACAACT-1
colData names(12): orig.ident nCount_Spatial ... spatial2 cell_type
reducedDimNames(0):
mainExpName: NULL
altExpNames(0):
Input Data Construction Start
Input Data Construction End
Start Marginal Fitting
Marginal Fitting End
Start Copula Fitting
Convert Residuals to Multivariate Gaussian
Converting End
Copula group 1 starts
Copula Fitting End
Start Parameter Extraction

I have been waiting for many days, but the program is just stuck here...

[dsong-lab]

Hi,

Thanks for your interst in scDesign3. Since you only test on 10 genes, would you mind trying n_cores = 1 and let me know if it works? Thanks!

[chthub]

@dsong-lab Thanks for your reply. It works. I am confused that when I need to adjust this parameter? Should I always use 1?

[dsong-lab]

Hi,
The number of cores you should use is a complicated problem; it depends on both your number of available cores and memory. If you use too many cores, the program will freeze quite often.

I would suggest: 1. The number of cores should be smaller than 1/10 of your total number of genes;
2. Make sure that each core can get >1 GB ROM;
3. I usually use <= 20 cores to make it safe.