Genome assembly with direct terminal repeats

[Phage Hunter]

Hi guys,

I'm having a bit of a problem with respect to viral genome assembly from my illumina data. Some of the genomes possess direct terminal repeats (DTRs) and this is causing all the assembly programs to assemble this into a pseudo-circular genome and then break this at some point. This rest of the genome appears to be fine and I've checked this with PCR, but I'm left with two halves of a gene at either ends of the molecule.

I've tried playing about with really large kmer sizes, but this produces no discernible effect as I suspect the DTRs are fairly large (~1000bp). I also mapped my reads back to the assemblies to try and find the actual end point and then resolve this by genome walking, but I could not see anything indicative of this. Unfortunately, as these are novel viral sequences, a reference guided assembler is out of the question.

Have you any experience with this? Strategies that deal with circular genomes should prove to be effective here.

Thanks in advance.

[awork]

Hello,
Have you found a solution to this problem? I am having the same issue when trying to assemble retrovirus genomes from illumina data. The LTR is causing the same issue you are reporting.
Thanks

[Phage Hunter]

Hi awork,

Yes, I found that using Geneious, a commercially available assembler solves most of my problems.

Geneious is good at assembling circular genomes and does this with my terminal repeats. You then get a linear contig with your genes in the correct order. If you wish to know the exact nature of your ends, then you can just use conventional Sanger sequencing on whole genomic DNA.

For the potential benefit of others: before getting geneious I had the idea of taking my misassembled genome and mapping the reads to this and viewing in software such as Integrated Genome Viewer. In theory (unless you removed these) there should be a number of short truncated reads mapped over the true ends. You could then break the contig at this point and rejoin to give a crude contig with genes in the correct order. This could perhaps be used as a reference genome in a reassembly (AlignGraph assembler could do this, or maybe the velvet columbus module) so as to give the true version of your genome.

I'd say the easiest option is Geneious. If you don't want to buy it, you could always get the trial version (15 days) and do as many assemblies as you can before this expires and then install on a different PC when you need it again.

With Kind Regards.

[Zapages]

I would suggest using MIRA in Geneious as well. Its really good with Viral genomes.

All the best with your project.