Keeping barcode information in bam file

[Jonathan Landry]

Dear all,

I have fastq files with barcode information in the header of each reads like:

@NS500188:164:H5WKHAFXX:1:11101:20350:1043 2:N:0:0AAGAGAGCAGA

Is there a way to keep this information (2nd field after space) in the bam file produced by STAR?

Thanks for your help,

Best,

Jonathan

[Felix Schlesinger]

Something similar to be bwa-mem -C option (http://bio-bwa.sourceforge.net/bwa.shtml) could be useful. Of course the more general way to handle this is to start alignment from (unaligned) BAM files, keeping all tags and other info.

[Alexander Dobin]

Hi Jonathan, Felix,

if you want to keep the barcode info in the header, you can simply remove the space between two fields, for instance:

awk 'NR%4==1 {print $1 "_" $2}'

or to keep just the barcode removing 7 first symbols from the 2nd field:

awk 'NR%4==1 {print $1 "_" substr($2,8)}'

To do it on the fly (wtihout creating intermediate files) you can add this command directly to --readFilesCommand.

Or if you have zipped files, you can make a script that you supply as --readFilesCommand /path/to/script.sh

zcat $1 | awk 'NR%4==1 {print $1 "_" $2}'

Felix,

it's not clear to me what output bwa-mem -C produces. Does it add a SAM attribute (tag) with the 2nd field of the name?

It should be easy to do - will add it to my list.

Cheers

Alex

[Felix Schlesinger]

Hi Alex,

yes it essentially takes everything after the space in the fastq and makes it a SAM tag. Avoids having to parse it out of the readname in the BAM later. But it is not very general, so I actually prefer aligning from BAM files as input and copying over all tags.

Felix

[Jonathan Landry]

Hi Alex, Felix,

Thanks for your answers.

Taking out the space could work nicely. I will give it a try.

Thanks a lot again.

Jonathan

[Alexander Dobin]

Hi Felix,

I have it in my plans to implement STAR input from BAM files, a few users have requested that.

Can Illumina pipeline output directly into BAM instead of FASTQ?

Cheers

Alex

[Felix Schlesinger]

There is no direct way to output unaligned BAMs directly instead of fastqs yet, but that will probably happen.

We already often only store BAM files long-term, so rerunning from BAM (instead of converting back to fastq) is already convenient, especially now with CRAM compression.

Felix