Filtering rRNA out using Bowtie2

[GWind]

Hello experienced colleagues,
I'm trying to filter out ribosome-related reads from my Illumina (Casava 1.8) pair-read (100 bp) data. The data was pre-filtered with Trimmomatic (adapters and low quality reads were removed). Now all this stuff exists in 3 files: “left.fq”, “right.fq” and “unpaired.fq”. Ribosome sequences are stored in the bowtie2-build “rRNA” base.

The simplest story about Bowtie2 syntax is in case of unpaired reads:

bowtie2 -p 8 --un /output/filt_unpaired.fq rRNA /input/unpaired.fq

It works.

But, more dramatic situation arises if I want to get filtered pair-reads in 2 separate files and additionally reads from broken during filtration pairs in a 3-d file.
I'm trying to do the following:

bowtie2 -p 8 -X 430 --un-conc /output/filt_pairs.fq rRNA -1 /input/left.fq -2 /input/right.fq

But, bowtie2 outputs just concordant pairs in 2 files... broken pairs are lost.
How to get them?

[caballien]

did you find a solution to this problem? I'm going to attempt to trim with trimmomatic and then filter rRNA reads with bowtie2. Were you able to keep both the paired and unpaired reads?

[boetsie]

Have a look at RiboPicker:

http://ribopicker.sourceforge.net/

You can download the tool and rRNA databases. It uses a modified version of bwa-sw for alignment against a rRNA database and splits your (paired) reads into rRNA and non-rRNA.

Boetsie

[Matt.Ralston]

Hello everyone. The --un-conc option outputs unconcordantly aligned pairs in two separate files, as you mention. The correct option for broken pairs is --un as before. include both flags in your statement

bowtie2 -p 8 -X 430 --un-conc /output/filt_pairs.fq --un /output/filt_single.fq rRNA -1 /input/left.fq -2 /input/right.fq

will output the filtered pairs in /output/filt_pairs.fq.1 and /output/filt_pairs.fq.2

and the single reads will be in /output/filt_single.fq

Hope this helps.
Cheers