Host Decontamination: bbTools or Samtools?

Question

[lotus28]

Hi,

I have always used Samtools to remove any human contaminants from my gut flora data. However, bbTools also allows you to do that. All the post I've seen mention using bbTools for small contaminants removal. Is it practical to use it for removing hg19 contaminants or should I stick with Bowtie + Samtools?

[WouterDeCoster]

I have always used Samtools to remove any human contaminants from my gut flora data.

Well, samtools has multiple subcommands, and you can do a lot with it. It is probably most helpful if you describe more in detail what you did (which commands you used).

[lotus28]

Basic samtools stuff: align against hg19 with bowtie2, extract reads with both ends unmatched, sort and convert filtered bams to fastq.

I was just wondering if I could remove one docker container from my pipeline by using bbTools for both quality trimming and decontamination

[GenoMax]

Yes you can do that.

Answer 1

[GenoMax]

Yes it is totally practical. @Brian Bushnell wrote a pipeline (which uses BBTools) to do that. You can find the post here. You could also use bbsplit.sh from BBMap tools and bin your reads in multiple ways.

[lotus28]

Yesterday I ran some tests to compare bowtie2 and bbTools decontamination. I used the same index provided in your link for both. bbTools takes longer to compute (15-30% longer) and deletes fewer reads. It also seems like bbTools has lower sensitivity, even in heavily contaminated runs. I think I will stick with my previous workflow.

[GenoMax]

I assume you tried the default options for everything. While you can stay with your pipeline I am sure there are options with bbtools that one can tweak to get results as good as what ever you are doing currently.

[lotus28]

Yes, probably tweaking bbTools pipeline might work, but I have not noticed any significant gain in speed to justify this tweaking. For the time being I will just stick with my current pipeline.