simulating a 2000cell*150000peak scATAC

[Chen-Li-17]

• It nearly spends a whole day to simulate, which is inconvenient.
• I encountered an error like this when I try to simulate a 2000cell*150000peak scATAC. Could you tell me how to fix it?

Input Data Construction Start

Warning message in asMethod(object):
“sparse->dense coercion: allocating vector of size 2.4 GiB”
Input Data Construction End

Start Marginal Fitting

Warning message in mclapply(seq_len(n), do_one, mc.preschedule = mc.preschedule, :
“scheduled cores 1, 2 did not deliver results, all values of the jobs will be affected”

Error in names(answer) <- dots[[1L]]: attempt to set an attribute on NULL
Traceback:

1. scdesign3(sce = sce_seurat, assay_use = "counts", celltype = "cell_type", 
 .     pseudotime = NULL, spatial = NULL, other_covariates = NULL, 
 .     mu_formula = "cell_type", sigma_formula = "1", family_use = "zip", 
 .     n_cores = 2, usebam = FALSE, corr_formula = "cell_type", 
 .     copula = "gaussian", DT = TRUE, pseudo_obs = FALSE, return_model = FALSE, 
 .     nonzerovar = FALSE)
2. fit_marginal(mu_formula = mu_formula, sigma_formula = sigma_formula, 
 .     n_cores = n_cores, data = input_data, family_use = family_use, 
 .     usebam = usebam, parallelization = parallelization, BPPARAM = BPPARAM)
3. suppressMessages(paraFunc(fit_model_func, gene = feature_names, 
 .     family_gene = family_use, mc.cores = n_cores, MoreArgs = list(dat_use = dat_cov, 
 .         mgcv_formula = mgcv_formula, mu_formula = mu_formula, 
 .         sigma_formula = sigma_formula, predictor = predictor, 
 .         count_mat = count_mat), SIMPLIFY = FALSE))
4. withCallingHandlers(expr, message = function(c) if (inherits(c, 
 .     classes)) tryInvokeRestart("muffleMessage"))
5. paraFunc(fit_model_func, gene = feature_names, family_gene = family_use, 
 .     mc.cores = n_cores, MoreArgs = list(dat_use = dat_cov, mgcv_formula = mgcv_formula, 
 .         mu_formula = mu_formula, sigma_formula = sigma_formula, 
 .         predictor = predictor, count_mat = count_mat), SIMPLIFY = FALSE)

[SONGDONGYUAN1994]

Hi Lee,
Thanks for your interest in our work! For the time complexity, it depends on your data dimension, model setting, and the number of cores. If your data is very high-dimensional, using more cores (e.g., > 10) can reduce the time dramatically.

For your error here, we seem to have some issues in the marginal regression model fitting. Please check two things:

1. Do you have any features with all 0?
2. If you set family_use = 'poisson', does it work? ZIP is usually less stable.

Without a reproducible case, sorry that I cannot say much about the reason. You can email me: dongyuansong@ucla.edu, and we can set a virtual meeting if it helps.

Best,
Dongyuan

[Chen-Li-17]

Dear Dongyuan,
Thank you for the reply, I'll rerun my code on your advice and give you feedback soon.