Is it possible to run CibersortX with a very small number of samples?

Question

[Aspire]

I am trying to impute group expression profile in CibersortX for a condition with 3 samples. For that, I have reduced my signature matrix to two cell types only (t-cells + "other").

However, running CibersortX (in GEP mode) throws an error

Error: $ operator is invalid for atomic vectors
In addition: Warning message:
In mclapply(1:no_cores, res, mc.cores = no_cores, mc.set.seed = FALSE,  :
  all scheduled cores encountered errors in user code
Execution halted

Adding samples removes the error.

Has anyone encountered this and succeeded to bypass?

[f2ff7a46]

Hi - Did you solve this issue, I am having the same problem

[Sarah]

also running into this!

Answer 1

[finch]

Hi! I had a very similar issue and it's likely not at all related to your number of samples but to the format of your mixture file! Cibersortx is super picky. Open your .txt file and make sure of a few things: 1. it's tab delimited 2. the first column is titled Gene_Symbols (if not add it and make sure to hit tab after to move over your sample names) and each sample has a unique name. 3. ensure that all gene symbols are capitalized and finally, 4. make sure there are no quotations. I was able to format my data in R pretty easily if that helps

[Aspire]

Thanks. I have checked that all files are tab delimited, with the first column named Gene_Symbols, with capitalized gene symbols, and no quotations. However the error still persists.

>Loaded 3 mixture samples, 26094 genes, and 2 cell subsets...
>Imputing representative cell type GEPs...done.

Errors were encountered in CIBERSORTx configuration, run aborted:

Error: $ operator is invalid for atomic vectors
In addition: Warning message:
In mclapply(1:no_cores, res, mc.cores = no_cores, mc.set.seed = FALSE,  :
  all scheduled cores encountered errors in user code
Execution halted

In my case (which seems to be different from your case), it has to do with the low ratio of samples:cell-types .

I say so, because when I enlarge the number of samples, the error is no more.

>Loaded 7 mixture samples, 26094 genes, and 2 cell subsets...
>Imputing representative cell type GEPs...done.
>Writing output to disk...done.
All done.