software for genome assisted scaffolding

[sasignor]

Hello all,
I have genome assemblies for two closely related species, one of which is much more complete than the other (and which I did not assemble). I am trying to improve my assembly using the information from the other assembly, but can't seem to find any programs that do this. I assembled contigs using abyss, and then improved the assembly using SSPACE. There is still a lot of room for improvement with the information from the other assembly.

Recommendations?

Thanks!

[Torst]

I have genome assemblies for two closely related species, one of which is much more complete than the other (and which I did not assemble). I am trying to improve my assembly using the information from the other assembly

Have you tried using Mauve's "Contig Mover" to align/order/orient your "bad" draft contigs against your "good" draft contigs?

URL: http://asap.ahabs.wisc.edu/mauve-ali...t-genomes.html

[sasignor]

I wasn't really looking for something that just reorders contigs, so much as uses an existing assembly to lower the threshold for joining contigs or to assist in the initial assembly.

[aaronrjex]

Columbus (which is an extension of Velvet: http://www.ebi.ac.uk/~zerbino/velvet/) seems to work quite well. Basically you align the reads from your query sequence to your reference sequence using a program like bwa, output a sam file and use this as the input for a Velvet assembly (the manual explains it all). I would suggest making sure your kmer_cov does not drop below 10 fold (Dan Zerbino does a much better job of explaining that one than I would be able to at http://www.ebi.ac.uk/~zerbino/velvet...th_choice.html).

Otherwise you could try the Sanger Institute's PAGIT pipeline (http://www.sanger.ac.uk/resources/software/pagit/), however that will probably only work if your reference assembly is of pretty high quality and you do not anticipate large rearrangements of the genome between the two sequences.

Hope that is of some use.

[sasignor]

Thanks!

My initial attempt used abyss to assemble the reads and SSPACE to improve the assembly. So far Columbus seems to be far, far worse than this which is odd. The input .sam file shows 93% of the reads being mapped to the reference but the output is much more fragmented than using abyss+SSPACE. It isn't a product of the reference as this is more complete. Perhaps I am not using the correct parameters.

./velveth my_dir 45 -reference genome.txt -shortPaired -sam sorted.sam

./velvetg my_dir -ins_length 350 -exp_cov auto -min_contig_lgth 300

PAGIT is still running, although it doesn't seem to do quite what I am looking for, but we'll see how it turns out.