Closed ncyue2 closed 11 months ago
Hi @ncyue2, sorry for the error. I am trying to fix it. Can I first make sure your inputs are raw counts (e.g., generated by htseq-count)? Your input file names may indicate they are normalized. DeMixT requires inputs are raw counts. Thanks.
Hi @jiyunmaths, thank you for the quick reply!
Unfortunately I do not have the raw counts, all I have access to is the tpm rate;
However, I converted the tpm rate to counts by reverse operation counts <- tpm*2600*10^{-6}
where 2600 is the average gene length; similar to the solution in this post
So my input data should be "imitating" the raw counts.
Extra: I estimates the proportion of stromal and immune components with scRNA-seq to be pi01 = 0.16
and pi02 = 0.63
respectively, but the same error occurs.
Thank you.
Hi @ncyue2, sorry I do not think it works to convert TPM to raw counts, since we lost the raw expression scale of each sample. Can I recommend to check if you can get access to the fastq or BAM file? If so, it will be much easier and more accurate to estimate the raw counts. Thanks.
Hi @jiyunmaths, that's okay! I will see if I can get access to the fastq or BAM file! Thank you!
Hi, I am following closely the tutorial on https://wwylab.github.io/DeMixT/tutorial.html, and the codes run perfectly fine.
But when i change the data to my own data, I received the error:
when i try to execute the line:
and I found out that this
res <- DeMixT(data_Y, data_N1, data_N2, niter = 5, nthread = 50, if.filter = TRUE)
only cause the error whenniter
is set to 5 or above.I have tried to set
filter.sd = 5
,ngene.selected.for.pi = 0
, orngene.selected.for.pi = 50
, but same error occurs. I am wondering what is causing the problem and any possible solution to the problems. Thank you.Below are my complete codes:
Other information: Operating system: Fedora Linux version: 6.4.12-200.fc38.x86_64 R version 4.3.1 (2023-06-16) Platform: x86_64-conda-linux-gnu (64-bit) other attached packages:
DeMixT_1.16.0
Thank you.