A stand-alone assembler based on the methods used for the Quip compression algrothim.
Pique is a pretty standard de Bruijn graph assembler, but it uses probabalistic data structures to achive extreme space efficiency, while remaining quite fast.
This is a research project, and at this point I do not recommend using it for any real analysis, but if you're curious about how efficient assembly can be, give it a try.
Download the latest tarball. Run ./configure && make install. Or, clone the
git repository and run autoreconf -i && ./configure && make install.
Running pique reads.fastq > output.mm will build a de Bruijn graph from the
reads, then output a sparse adjacency matrix in Matrix Market
Exchange format.
There are a number of options which you can read about with pique --help.
Most importantly -t T will run pique concurrently on T, threads, and -k K
controls the k-mer size of the de Bruijn graph.