Eliminate rRNA contamination with Bowtie

[alisrpp]

Hi,

I'm working with Illumina 101bp reads of marine sponges. Before trimming the adapters and filtering by quality with the Trimmomatic i run FastQC to check, among other things, the amount of overrepresented reads to have an idea of the rRNA contamination of my raw reads. I also run FastQC after the Trimmomatic to check exactly the same thing.

Now i'm dealing with the rRNA contamination. I'm aligning my reads against all the Porifera rRNA sequences in NCBI using Bowtie with the parameter --un that keeps the unaligned reads. My logic made me do that with my already trimmed and filtered reads but maybe is better if i use as an input my 2 FASTQ paired-end (/1 and /2) raw files (without any trimming or filtering).

Also the script that i'm using is this one:

$ bowtie -a --best --strata -t --al aligned.fq --un unaligned.fq Porifera_rRNA <nameofmyfile>

I'm completely new using Bowtie so i don't know if maybe for what i'm trying to do is better to use some other parameters.

I'm looking for a little bit of feedback, i'm the only person in my lab working with that kind of stuff.
Anyone out there can help me?

Thanks.

[westerman]

Personally I like bowtie2 -- the parameters can be either individually tweaked or globally tweaked via '--fast-local', '--sensitive', etc.

[cascoamarillo]

I think your approach with bowtie1 is perfectly reasonable; first clean a little all the reads (low Q, adapters...) and then filter out rRNA reads with your Porifera_rRNA database. I'm not sure about using the parameter -a/--all (instructs bowtie to report all valid alignments) along with the --un option (write all reads that could not be aligned to a file). The --best option would be work fine for cleaning rRNA reads, don't think the --strata would be necessary; you will get another fastq/fasta file with the filtered reads.