[DAS] Poka: alpha test of UCSC-based DAS2 server

[Gregg Helt]

I just deployed an alpha test of Poka, my UCSC-based DAS2 server onto the
Amazon Web Services EC2 cloud.  You can try it out at:
     http://www.genomancer.org:9095/das2/ucsc/genome/sources

Some sample queries in addition to the sources query above:

Segments (latest human genome):
http://www.genomancer.org:9095/das2/ucsc/genome/hg18/segments

Types (latest human genome):
http://www.genomancer.org:9095/das2/ucsc/genome/hg18/types

Features (latest human genome, genscan predictions around GABPA & ATP5J
region):
http://www.genomancer.org:9095/das2/ucsc/genome/hg18/features?segment=http%3A%2F%2Fwww.genomancer.org%3A9095%2Fdas2%2Fucsc%2Fgenome%2Fhg18%2Fchr21;overlaps=26022853%3A26067933;type=http%3A%2F%2Fwww.genomancer.org%3A9095%2Fdas2%2Fucsc%2Fgenome%2Fhg18%2Fgenscan

Features (latest human genome, mRNA alignments around GABPA & ATP5J region):

http://www.genomancer.org:9095/das2/ucsc/genome/hg18/features?segment=http%3A%2F%2Fwww.genomancer.org%3A9095%2Fdas2%2Fucsc%2Fgenome%2Fhg18%2Fchr21;overlaps=26022853%3A26067933;type=http%3A%2F%2Fwww.genomancer.org%3A9095%2Fdas2%2Fucsc%2Fgenome%2Fhg18%2Fmrna

If you want to try this out with a DAS2 client, you can use IGB and add the
following line to your igb_prefs.xml file:
<das2_server name="genomancer UCSC poka" url="
http://www.genomancer.org:9095/das2/ucsc/genome/sources" />
IGB has a bug that leads to duplicating each pairwise alignment annotation
(like the mRNA alignment example above).  This has been fixed in the source
code repository, but until there's a new IGB release it will show up if
you're using the standard IGB WebStart distribution.

This alpha test supports annotation types that are represented in the UCSC
genome database as BED, GENEPRED, or PSL style tables.  This covers roughly
half of all the annotation tracks available in the UCSC genome database.
I'm working on support for the other half...

Also, this version doesn't use alternative content formats or a number of
other optimizations, so response time is a bit slow.

And for this test phase, since it's using UCSC's shared public MySQL server,
I've restricted the server to only return features if there are < 1000
"top-level" features (for the mRNA example above, this averages out to
~10,000 total features including exons).  Otherwise it returns a
REQUEST_ENTITY_TOO_LARGE error as per the DAS2 spec.  Keep this in mind if
you're trying Poka out in IGB -- other than a diagnostic message in the
output console, IGB will silently ignore the error but no annotations will
show up.

Poka is implemented as a plugin to the Trellis DAS2 server framework.  Hence
the name Poka -- my favorite trellis climber is the passion flower vine, and
banana poka is a species of passion flower that has a tenuous UCSC
connection via sharing part of its name with the official UCSC mascot (the
banana slug).

If you get a chance to try it out, please let me know what you think.

       Gregg