Add processing of sample names consisting only of numbers

[ValentineObrezanenko]

Description of feature

Hi.
I faced the problem that if I have only numbers written to the sample ID, then in some scripts she starts counting them as numbers. And there is also a comparison of a number with a string, which causes an error. I think it's worth paying attention to this. For example, the run_drimseq_filter script.R and suppa_split_file. You can take this into account in the line samps$sample_id < - as.character(samps$sample_id).

[gianfilippo]

Hi,

similar issue here. My samples IDs start with a number and end with a character. The run_drimseq_filter.R fails with the following error
Error in DRIMSeq::dmDSdata(counts = counts, samples = samps) :
all(samples$sample_id %in% colnames(counts)) is not TRUE
Calls: -> stopifnot
Execution halted

This is because after the salmon count file is read, a count data.frame is created
counts <- data.frame(gene_id = tx2gene$gene_id, feature_id = tx2gene$tx,cts)

and at this point the sample IDs are not handled properly and. an "X" is prepended, eventually resulting in a mismatch

I added "check.names = F" to fix this, but I am not sure this is the best way.

Thanks

[gianfilippo]

UPDATE
after the previous correction, a similar error
Error: all(samples$sample %in% colnames(counts)) is not TRUE
is called.

This is originally caused by the following call
d <- DRIMSeq::dmDSdata(counts = counts, samples = samps)

again, this is because the DRIMSeq::dmDSdata function does not handle colnames starting with numbers and prepends an "X". This results in a count matrix saved with colnames not matching sample names.

At the moment I added the following line
colnames(d.counts) = colnames(counts)

after this bit of code

Take count data

d.counts <- counts(d)

This should be ok, since the two DRIMSeq calls, dmDSdata and dmFilter, do not filter out samples, only features, as far as I understand.

The job completed successfully.