readAllelicCountsFile() can be very slow (with suggested fix)

[luyh-xp]

I was running PureCN on results generated by the GATK4 Somatic CNV workflow upstream, and I notice that it takes hours for readAllelicCountsFile() to parse an allelic counts tsv file. Admittedly my tsv file is fairly large (~2.4M lines), but this is still way too time-consuming.

It turns out that most of the time is spent on a single line of code (readAllelicCountsFile.R, line 79):

alt(vcf) <- DNAStringSetList(lapply(inputCounts$ALT_NUCLEOTIDE, DNAStringSet))

As far as I understand it, lapply() is basically equivalent to the native R for loop and it is not very efficient. So I replace it with the following line:

alt(vcf) <- DNAStringSetList( split( inputCounts$ALT_NUCLEOTIDE, 1:length(vcf) ) )

And it drastically reduces the running time of this function to a few minutes. Therefore, I recommend the developers to incorporate this fix so that future users won't have to solve the same problem when they are running PureCN on GATK4 outputs. Thank you very much for maintaining this amazing project!

[lima1]

Thanks @luyh-xp . I'm sorry, I never really used that code myself.

Would this even work?

alt(vcf) <- DNAStringSetList(as.list(inputCounts$ALT_NUCLEOTIDE))