the results of fit_marginal are all null

[Santanevrgivup]

my code: data <- construct_data( sce = sce, assay_use = "counts", celltype = "cell_type", pseudotime = NULL, spatial = c("spatial1", "spatial2"), other_covariates = NULL, corr_by = "1" ) marginal <- fit_marginal( data = data, predictor = "gene", mu_formula = "s(spatial1, spatial2, bs = 'gp', k= 50)", sigma_formula = "1", family_use = "nb", n_cores = 10, usebam = FALSE ) origin data is from tech DBiT-seq, and with dim(1988*1662) after preprocessed.

here is the output of fit_marginal function

so in the next step(fit_copula) also report errors like this : Error in colnames<-(*tmp*, value = rownames(sce)) : attempt to set 'colnames' on an object with less than two dimensions Calls: fit_copula -> convert_n -> colnames<- In addition: Warning message: In mclapply(seq_len(n), do_one, mc.preschedule = mc.preschedule, : all scheduled cores encountered errors in user code Execution halted I would greatly appreciate your help！

[SONGDONGYUAN1994]

Hi, The result means that the fit_marginal fails for all the genes. This is usually caused by some errors in the input data (output of construct_data). Could you set all n_cores = 1 and check the error message?

Best, Dongyuan

[Santanevrgivup]

Dear Dongyuan, Following your suggestion, I set all n.cores values to 1 and checked the output of the construct_data function. I couldn't find any anomalies. So the output of fit_marginal still results in all empty values, and this process didn't produce any error messages. Could you please provide me with some further guidance? I greatly appreciate your help.

[SONGDONGYUAN1994]

This is weird. Could you share your sce data with me? You can mask the gene names if you are worried about data leaking. I can check it.

Best, Dongyuan

[Santanevrgivup]

Of course, it would be my pleasure！This data is public, so there are no concerns about data leakage. Please feel free to proceed without any worries.This compressed file contains three files: 'input_sce' is the sce_obj that has been preprocessed and is ready to be input into construct_data, 'code.R' contains part of the relevant code, and 'GSM418~' is the raw data. Thank you again! Santa Check.zip

[SONGDONGYUAN1994]

@Santanevrgivup Hi Santa,

I found the problem. Please convert your spatial locations (spatial1, spatial2) from chr to numeric. Now they are characheters so that the model cannot fit.

Best, Dongyuan

[Santanevrgivup]

Hi Dongyuan,

The problem has been solved by following your suggestion! Thank you once again for your patient guidance. Sincerely wish you good health and lasting academic vigor!

Santa