reorder contigs given by nucmer

Question

[Phil S.]

Hi there,

I used nucmer to place (de novo assembled) contigs onto a close relative and as far as I can tell from the dotplot it turned out really really well. Now I want to reorder the contigs into the ordering found by nucmer. Is there any tool or script out there which is able to do that?

Thanks,
phil

Answer 1

[SES]

There is a tool called ABACUS that is designed for that exact purpose, and in my experience it works really well. This program actually uses nucmer to align the contigs to the reference, so this sounds like a good fit. ABACUS is also a part of the larger PAGIT pipeline for assembly improvement.

[Phil S.]

Hi,

thanks for the fast reply. Unfortunatelly it seems like the abacas does not get quite the results using nucmer alone. I dont know why but using nucmer alone I end up getting something like this:

Using abacas I get just around half of the bases placed and the rest is filled up with Ns. I can't see any switch for the amount of Ns inserted or when they are inserted etc...

edit: I just saw that MAUVE also has a 'move contig' option which is able to do that. Will keep you posted if that worked better. Anyways if someone has a different solution I'm more than happy to hear about it!

Cheers

[SES]

Have you tried modifying the options to ABACUS? I would check your settings for running mummer alone, and then give those settings to ABACUS. You can change the percent identity, coverage, etc. of matches, so you should be able to get close to the same results. There will likely be differences because of the post-processing of the matches.

[Phil S.]

I have a look into that right now. Thanks!

[Prakki Rama]

@SES But ABACAS does not stitch the tips which are overlapping. It just reorders and places N's in between contigs. In that case, there will be same region twice in a scaffold. Isn't it?

[Phil S.]

Yep it seems like ABACAS does not stitch. I modified ABACAS that it nearly produced the same results. Though I am figuring out how to stitch manually....(meanwhile testing the 'move contigs' function MAUVE provides..)

Thanks for all your thoughts / ideas

Answer 2

[Andreas]

Hi,

I'm a bit late to the party but:

Mummer's show-tiling does the trick. It can also generate a concatenated pseudo-sequence (gaps filled with Ns) if you use option -p

Andreas