finding repeats in de-novo assembled contig from PacBio reads

Question

[aindap]

Dear BioStars Community:

I performed a de-novo assembly of PacBio reads using Canu for a viral genome. I have my resulting unitig from the canu pipeline. I am now interested in characterizing repeats in my resulting assembly. I'm new to assembly and repeat identification. One simple approach was taking the reads used to form the assembly, align them against the assembly with MUMmer, and take a look at the resulting dot plot? Are there any more sophisticated approaches that would yield better results?

[reza]

why you do not use Tandem Repeat Finder or RepeatMasker to do this?

[Brian Bushnell]

Assuming the assembly is correct, it seems to make more sense to align the assembly to itself rather than aligning reads to the assembly.

Answer 1

[arnstrm]

Yes, you should probably try RepeatModeler, which can detect repeat families (de novo) and classify them. It has worked well for both model/non-model species and is very easy to run (it does have few dependencies to install though: TRF, RECON, RepeatScount, NSEG).

EDIT: you can find my sample run script here!