Can I use umi_tools to discard cell barcodes and extract the UMI while keeping the UMI in the read?

Question

[cag104]

I have paired end reads where the read structure is as follows:

>Read1
5'-6N-20CB-XXXXXXXXXX....XXXXGS-3'

The 6N corresponds to the UMI, the 20CB corresponds to 20 basepairs of cell barcode and X..XGS corresponds to genomic sequence.

I want to extract the UMIs and place them in the header, but NOT discard them (reattach them to the read similarly to how umi_tools treats X's) and discard the 20bp of cell barcode so that the read structure becomes:

>Read1::NNNNNN
5'-6N-XXXXXXXXXX....XXXXGS-3'

Is this possible to do with umi_tools? If not, what would you suggest?

[i.sudbery]

Do you mean that you don't want the cell barcode to be transffered to the read header?

Answer 1

[i.sudbery]

You were happy to transfer the CB to the read header as well, then you could do:

$ umi_tools extract -I read1.fastq.gz --read2-in=read2.fastq.gz --stdout=read1.cb.fastq.gz --read2-out=read2.cb.fastq.gz --bc-pattern='^.{6}(?P<umi_1>.{20}' --extract-method=regex  --log=cb.log

$ umi_tools extract -I read1.cb.fastq.gz --read2-in=read1.cb.fastq.gz --stdout=/dev/null --read2-out=read1.umi.fastq.gz --bc-pattern=NNNNNN --log=read1.umi.log

$ umi_tools extract -I read1.cb.fastq.gz --read2-in=read2.cb.fastq.gz --stdout=/dev/null --read2-out=read2.umi.fastq.gz --bc-pattern=NNNNNN --log=read2.umi.log

Obviously if you only have 1 read you can skip the final step, and the read2 bits in the first step.