How To Run Abyss On One Chromosome Only ?

Question

[madkitty]

I want to do a de novo assembly on the mitochondrial chromosome only with ABySS.

I don't see any option that can tell the software to do a denovo assembly on only one chromosome (in this case, mitochondrial chromosome).

If like me you don't know about ABySS I might as well use SOAPdenovo.. but the problem is the same, I have no clue how to start the assembly for one chromosome only.

Answer 1

[Mikael Huss]

By definition, de novo assembly is done "blindly", without using a reference such as a mitochondrial genome. Perhaps you are thinking of reference-guided assembly such as done by e g Cufflinks or Scripture, where a reference sequence is used as support.

If you don't have a reference sequence and want to do a de novo assembly, just run ABySS on all your reads, and BLAST the resulting contigs against NR or some subset thereof. This should allow you to identify which contig(s) correspond(s) to the mitochondrial genome.

[madkitty]

Thanks for your answer. We have a whole genome sequence of a mouse thou ..Do you know how long it would take approximatively ? Something like 2-5 days I guess ?

[Mikael Huss]

It really depends on the number of reads and your computing setup (sorry I can't be more specific). If it's mouse, you could do a reference-guided assembly as mentioned in the answer and that should be much faster than de novo. But perhaps you have specific reasons to do it de novo?

[madkitty]

I'm looking to do a denovo alignment of the mitochondrial DNA. I'm a Biochemist, not a bioinformatician so I understand the concept of a reference-guided assembly but I have no clue how to do that.. And the readme file of ABySS and SOAPdenovo doesn't show any specific option to do that.

[Leszek]

SOAPdenovo is for de novo assembly only (as name suggest;) ) and designed to work with Illumina PE reads.

Answer 2

[Leszek]

Or, align your reads on mitochonrion, select only reads that align and build assembly with those. Haven't tried it myself, but in principle it should work.

This should save you some days of computation;)

Edit:
First, build index of your genome (bowtie-build), then align reads. You can output aligned reads to mit.fq (--al mit.fq) and use this file for de novo mitochondrion assembly. In case, you have paired end reads, you can align them separately and then parse output fq file to make sure both pairs are present in output fq file.
I recommend increasing number of mismatches (-n3) and shorten seed length (-l21). It should be more sensitive and still very fast.

bowtie-build mit.fa mit.fa
bowtie --al mit.fq -S -n3 -l21 -q mit.fa reads.fq > algs.sam

[madkitty]

So can I select reads from the BWA file after alignment ? Then do a denovo align with ABySS ?

[Leszek]

I recommend bowtie, as it output aligned reads, is faster and you don't care that much about specificity here.

[madkitty]

Thanks I'll try this out Monday :)

[madkitty]

Hi there, I've tried bowtie but I got an error message when I run bowtie-build. It says "Error: could not open mit.fa"

Also is 'reads.fq' the path to my fastaq file from the sequencer ?

[Leszek]

yes, and mit.fa is you reference mitochondrial genome, so change it accordingly

[madkitty]

So should I provide the reference mitochondrial genome as well ?