scFEA output

[niuruize]

For the same data, when I work with a smaller number of cells (1000), it works fine. But when I'm working with a lot more cells (>50000),The flux and balance files are NA.Why is that？

[ZZZhuLF]

I also encountered the same situation, someone can answer?

[niuruize]

More cells can be analyzed after filtering the data（nFeature_RNA > 500). I've analyzed 100,000 cells at a time. But as you have more cells, it is advisable to break the data into parts.

[ZZZhuLF]

Thank you for your reply. Is it because tens of thousands of cells are too many? Then I will try to take subsets by cell type

[niuruize]

I suspect the effect of low quality cells

[ZZZhuLF]

nohup python src/scFEA.py --data_dir data --input_dir input
--test_file epmn.csv
--moduleGene_file module_gene_m168.csv
--stoichiometry_matrix cmMat_c70_m168.csv
--cName_file cName_c70_m168.csv
--sc_imputation True
--res_dir outputmn
--output_flux_file outputmn/f_flux.csv
--output_balance_file outputmn/f_balance.csv&

I have tried many methods, and the matrix obtained is all NA, including not limited to taking subsets, using the whole set, and removing low UMI cells. The data I am using at present is human data. Last year, I used scFEA to analyze the data of many mice, and it was no problem

[tobylanser]

Hi @niuruize @ZZZhuLF ,

I also encountered this problem. I was able to get around it by using the whole gene expression matrix (i.e. not only HVGs, etc). Make sure that whatever matrix you're exporting contains all (or most) genes. I suspect the the outputs contains all NAs because the genes in the expression matrix could not be matched to the gene modules.

[LiuCanidk]

Hi @niuruize , I have some questions on the out-of-memory error with big datasets.
As you mentioned above, big data should be broken into small parts. So do you mean this do not influence the result? That is, scFEA works seperately for each cell's transcriptome, and it won't change even when seperated to individual cells?

[Lanlinn]

I encountered the same problem. And after I removed the cells that did not express any of the metabolism-related genes (genes in the module_gene file) , I was able to run the analysis normally.