Removing contaminants (bacteria, phage) in genomic dna sequences

[NGS_New_User]

Hi Everyone,
I have dna, paired-end genomic sequences that I want to perform de novo assembly on, but I want to 'clean' them up first ie remove contaminants, before de novo assembly.
I have already done the trimming out of low quality reads and adapter removal. I would like to preferably map the reads or blast them to a bacterial and/or phage database, then keep the un mapped reads for further analysis downstream.

Could any one please guide me on how I can approach this?

I was thinking of downloading the bacterial and/or phage genomes to my local computer, but then there are 1000's of genomes presently.

Ideas and suggestions will be, as always, very appreciated!

[dsenalik]

I used DeconSeq for this, but I had to install and run it locally, since my job was in the queue for two weeks waiting and then vanished. (this was back in November).

I was lucky, only a very tiny trace of fungal+bacterial ribosomal genes. Though I was glad to see those, it indicated that the filter worked.

But, of course you can only check for contamination of sequenced organisms in the Deconseq database. I would be curious if there could be a more general filter, but I can't really see how, for a newly assembled genome.

Edit: Oh, and if you do this with assembled contigs, instead of reads, realize that the entire sequence is flagged as contaminated or not. I chopped up my assembled sequence into scaftigs prior to running (does anyone else ever use the term scaftigs?)

[Brian Bushnell]

You could BLAST a few thousand reads, see what they hit, and then just download the genomes of organisms that appear to be contaminants for your filtering. If you have a reference of your target genome, map to the reference, then only BLASTthe unmapped reads.

Once you have references, bbduk or bbmap can decontaminate using kmers or mapping, respectively, with the "outu" (output unmapped/unmatched) stream.

[cliffbeall]

My approach has been to do the assembly first, then try to remove contaminant contigs. This greatly reduces the amount of computation that has to be done. Chimeric contigs of contaminant and target should be extremely rare.

If you have a microbial genome, IMG has some tools for finding contamination, eg:

https://img.jgi.doe.gov/er/doc/SingleCellDataDecontamination.pdf