When using an Illumina Samplesheet with Adapters specified in the [Settings] section, bcl2fastq, cellranger mkfastq or BCLconvert automatically perform adapter trimming, e.g.
I do not think we want to use this feature because
trimming is done in a separate step using fastp by default, which would result in double-trimming (fastp also reliably auto-detects the adapters, there is no need for storing the adapter sequences).
trimming might result in issus with single-cell or UMI data. In fact 10x urgently recommends to disable trimming during demux:
I therefore suggest to include a step that removes those lines from the input samplesheets, unless it is specifically disabled via a config flag. This can be achieved e.g. in groovy with a snippet such as
samplesheet_in
.readLines()
.each { line ->
if ( line =~ /Adapter,[ACGT]+,/ ) {
line = line.replaceAll(/Adapter,[ACGT]+,/, 'Adapter,,')
}
else if ( line =~ /AdapterRead2,[ACGT]+,/ ) {
line = line.replaceAll(/AdapterRead2,[ACGT]+,/, 'AdapterRead2,,')
}
samplesheet_out << line + '\n'
}
A section should be added to the documentation dedicated to trimming.
Description of feature
When using an Illumina Samplesheet with
Adapters
specified in the[Settings]
section, bcl2fastq, cellranger mkfastq or BCLconvert automatically perform adapter trimming, e.g.I do not think we want to use this feature because
fastp
by default, which would result in double-trimming (fastp also reliably auto-detects the adapters, there is no need for storing the adapter sequences).I therefore suggest to include a step that removes those lines from the input samplesheets, unless it is specifically disabled via a config flag. This can be achieved e.g. in groovy with a snippet such as
A section should be added to the documentation dedicated to trimming.
CC @apeltzer