Open Source Java De Novo Assembly Software

Question

[lyz10302012]

Is there any open source java de novo assembly softwares or de bruijn graph algorithms?

Thanks

[zam.iqbal.genome]

Do you mean stuff that actually works on real data, or just for information/investigation?

[lyz10302012]

I mean stuff that actually works on real data.

[zam.iqbal.genome]

Ah. Well in that case I have never heard of one I'm afraid. Genome assembly is quite memory hungry, and so I think people tend to want to maintain very tight control of how they allocate and release memory. I could be wrong, there have been a lot of assemblers written, and you might find one that works on bacterial data - but if so I expect someone will reply here to tell us. best Zam

[lh3]

I agree. De novo assembly is the most memory demanding application in computational biology, while Java is notorious for its huge memory consumption. We can hardly see how they fit together...

Answer 1

[Rayan Chikhi]

There exists one open-source Java software that constructs the de Bruijn graph (but it does not assemble). http://grafia.cs.ucsb.edu/msp/download.html

There does not appear to exist any open source de novo assembler written in Java.

Source: http://seqanswers.com/wiki/Special:BrowseData/Bioinformatics_application?Language=Java&Bioinformatics_method=Assembly&Biological_domain=De-novo_assembly

EDIT:

Indeed, Contrail can be considered as a de novo assembler written in Java.

Answer 2

[Chris Whelan]

This is probably not exactly what you are looking for, but Michael Shatz's group is working on a Hadoop-based de Bruijn graph assembler called Contrail:

http://sourceforge.net/apps/mediawiki/contrail-bio/index.php?title=Contrail

Since it is a native Hadoop application it is mostly written in Java, although it appears that they also invoke other non-Java programs (FLASH and QUAKE) as part of their workflow.

This gets around the memory management features of Java that make a traditional single-machine algorithm hard to implement by using Hadoop's mechanisms of streaming data to and from disks across a cluster.

I am not sure what its state of readiness/current ability to run on real data is.