announcement: ABySS 1.9.0 release

[Ben Vandervalk]

Hi everyone,

I am pleased announce the release of ABySS 1.9.0.

Release page: http://www.bcgsc.ca/platform/bioinfo/software/abyss/releases/1.9.0
Source tarball: https://github.com/bcgsc/abyss/releases/download/1.9.0/abyss-1.9.0.tar.gz

Description:

This release introduces a new paired de Bruijn graph mode for assembly. In paired de Bruijn graph mode, ordinary k-mers are replaced by k-mer pairs, where each k-mer pair is separated by a fixed-size gap. The primary advantage of paired de Bruijn graph mode is that the span of a k-mer pair can be arbitrarily wide without consuming additional memory, and thus provides improved scalability for assemblies of long sequencing reads.

This release also introduces a new tool called Sealer for closing scaffold gaps, new Konnector functionality for producing long pseudo-reads, and support for the DIDA (Distributed Indexing Disptached Alignment) parallel alignment framework.

- Ben

[Son Pham]

Thanks Ben, This is an amazing work. In the new benchmark of Abyss, did you see N50 improvement of the new paired de bruijin graphs comparing to the old de Bruijn graph implementation?

-Son, 

--
You received this message because you are subscribed to the Google Groups "ABySS" group.
To unsubscribe from this group and stop receiving emails from it, send an email to abyss-users...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Ben Vandervalk]

Hi Son,

When comparing assemblies where paired k-mer span = k-mer size in standard de Bruijn graph mode, the N50 is generally the same or slightly worse, depending on how large you set the size of the individual k-mers in the k-mer pair.  This makes sense because you are discarding some sequence information in the gap between k-mer pairs.

However, where we expect the paired de Bruijn graph to start showing its value is in situations is when you have long high-quality sequencing reads (e.g. Moleculo, 10X genomics) and you want to use a k-mer size that would be impossible with standard de Bruijn graph due to memory limitations.  So far, there aren't very many data sets available in that category.

- Ben

[Son Pham]

Ben, 
Thank you Ben --- Now I understand the value of this approach in long reads! 

4 years ago when we work with Pavel&Paul on the paired de Bruijn graphs, we were aiming for read pairs, but now I see its value in the coming long read era.

Thank you!

-Son.  

[Ben Vandervalk]

Cool :-) We'd be curious to hear about the results, when you get there!

- Ben

On Fri, May 29, 2015 at 7:44 PM, Rick Westerman <west...@purdue.edu> wrote:

As soon as my sysadmin installs 1.9.0 I’ll have a data set to try.  Bird.  1.2 GB estimated genome.  Moleculo long reads improved the assembly just a bit but did not seem to stitch together the pair-end contigs.

--

Rick Westerman

west...@purdue.edu