kdzimm
commented
2 years ago @parkjooyoung99 are there any individuals among your data that have very few high quality cells? Due to zero inflation - I've noticed with some other datasets - that you always need more cells than expected. Upwards of ~50 per sample.
Hello, I tried running two commands with my own single cell data. Since 'filter_counts' and 'compute_data_summaries' did not have any problems, I am kind of sure that my input file is not the matter.
Following source codes from [https://rdrr.io/github/kdzimm/hierarchicell/man/error_hierarchicell.html], I found out that there were no data that fits "genecounts <- genecounts[which(apply(genecounts, 1, mean) > 5), ]" so I just changed it as 0.
Than, I met another problem at "zlmCond <- suppressMessages(MAST::zlm(~ ngeneson + Status + (1 | DonorID), sca, method='glmer',ebayes = F, strictConvergence = FALSE)" that giving Error message of the following Completed [----------------------------------------------] 0% with 0 failuresError : number of levels of each grouping factor must be < number of observations (problems: DonorID) Completed [----------------------------------------------] 1% with 0 failuresError : number of levels of each grouping factor must be < number of observations (problems: DonorID) Completed [>---------------------------------------------] 1% with 0 failuresError : grouping factors must have > 1 sampled level Completed [>---------------------------------------------] 2% with 0 failuresfixed-effect model matrix is rank deficient so dropping 1 column / coefficient Completed [>---------------------------------------------] 3% with 0 failuresError in asMethod(object) : not a positive definite matrix
I have no idea what the numbers of observations is cause the DonorID has 6 levels and I am quite sure that objects of sca will be larger than 6.
Just in case I captured part of my data,
If more information is needed, please let me know. Thank you!