Run resolveMutSignatures for a single sample

[Eugene79]

I tried to run resolveMutSignatures for a Mutation Counts object with one sample. Error message:

Error in apply(mutCountData, 2, (function(clmn) { :
dim(X) must have a positive length

I assume that initially resolveMutSignatures was supposed to run on multiple samples. Is there a way to run it on a single sample?

[dami82]

Sorry for the late reply and thanks for reporting this issue. Yes, the analysis isn’t really meaningful unless you have at least few samples. The idea is to identify the basic common components (Signatures) found in a series of samples. If you only have one sample, you can still extract signatures, but the results are unreliable. However, the exception should be handled more gracefully. Different story if you are just deconvoluting using known signatures, that is totally fine and shall work. I am a bit swamped these days, but I’ll look into this as soon as I can. Best regards, Damiano —— Damiano Fantini Sent from my iPhone

…

On Oct 26, 2020, at 08:17, Eugene79 ***@***.***> wrote:  I tried to run resolveMutSignatures for a Mutation Counts object with one sample. Error message: Error in apply(mutCountData, 2, (function(clmn) { : dim(X) must have a positive length I assume that initially resolveMutSignatures was supposed to run on multiple samples. Is there a way to run it on a single sample? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

[eibol1]

Hi,

I got a similar error. I have 23 VCF files, one per sample. I tried to run the MutSignature in one sample and got:

Error in decipherMutationalProcesses(input = blca.counts, params = blca.params) : mutSignatures does not allow to extract mutational signatures from a single sample. Please, add more samples and run again!

How Can I add more samples?

How can I merge these VCF files into one VCF so MutSignature can work? If this isn't possible, is there any way to merge all calls and differentiate them by sample name with importVCFfiles() function?

[dami82]

Hello.

Thanks for using mutSignatures and for reaching out with your question.
Yes, mutSignatures does NOT allow users to extract signatures from a
single sample. So, you need to pass a mutationCounts object with 2 or
(possibly) more samples.
Also, I am doing my tests with the latest version of the library, and at
this moment that is ver 2.1.5.

What you need to do is to pass a character vector including a list of
existing VCF files to the importVCFfiles() function. This will
automatically assign the appropriate VCF file name (stripped of the paths)
to each corresponding mutation. You can edit the file name later on (for
example via dplyr::mutate). Below I attach an example. Let the folder
named ~/VCF_fold/ be a directory where you have your gz-compressed VCF
files (.vcf.gz). This code looks for those files and builds a vector of
file paths that you can feed into importVCFfiles().

# import mutation data from a list of VCF files
vcf_ext <- '\\.vcf\\.gz$'
de_novo_fld <- '~/VCF_fold/'
my_vcfs <- grep(pattern = vcf_ext, dir(de_novo_fld), value = TRUE)
my_vcfs <- paste0(de_novo_fld, my_vcfs)
head(my_vcfs)

# Read a list of VCF files via importVCFfiles()
x <- importVCFfiles(vcfFiles = my_vcfs)

# Check data.frame dims
dim(x)

Now, you should see the filenames in the column named SAMPLEID. THe x
object is just a data frame at this point, so feel free to do a left join,
or use regex, or edit as you wish.

x <- x %>%
  dplyr::mutate(SAMPLEID = gsub('[[:punct:]]', '_', toupper(.$SAMPLEID)))

From here, you can proceed as you already did.
I hope this is useful.

Best regards,
Damiano

[eibol1]

Thanks @dami82, it worked well!!