Biolite

=====

* Fixed a problem with GNU parallel sometimes scrambling lines of output on
Ubuntu. (13)
* Fixed errors in 'sra import' when importing Agalma test data sets. (11)
* New '--metadata' option for 'sra import' only downloads metadata from SRA
without downloading the full data set. (11)
* Fixed bug in 'catalog insert' where unspecified fields were overwritten on
updates. (11)
* Improved robustness of the Butterfly stage in the Trinity workflow (10):
o Replaced multijar with GNU parallel for concurrent execution of
Butterfly commands.
o Added prefiltering of Chrysalis components to remove ones that are smaller
than the minimum transcript length.
o Reduced the number of threads used by half, since each Butterfly command
uses around 150-200% CPU.
o Switched from the raw 'butterfly_commands' output of Chrysalis to the
adjusted 'butterfly_commands.adj' which includes additional Butterfly flags.
* Improved the 'diagnostics delete' command so that it can handle incomplete
runs, and added an option '--incomplete' to 'diagnostics list' to list only
incomplete runs. (8)

=====

* Added wrappers for SGA and CGAL.
* Updated Trinity version to r2013-02-25.
* New SRA import/export tool. Import is tested and working. Export creates XML
submission files that validate against the SRA schema, but we are still
working out problems with the automated submission process.
* Reports can include javascript sources for D3js and jsPhyloSVG, and use a
new template with improved CSS styling.
* New 'diagnostics run --commands' dumps list of commands.

=====

* Fixed bug with duplicate IDs showing up in annotated FASTA headers
(from workflows.blast_annotate_seqs).
* Pipelines now create their own scratch directory in the current working
directory, using the format "[name]-[runid]". Restarts must be run from
inside that subdirectory (where the .chk file is located).
* Rewrote treepruning to fix errors with recursion and non-deterministic
output in paralogy pruning. At each iteration, paralogy pruning now
simaultaneously clips and outputs each maximally inclusive subtree that
includes no more than one tip for each taxon.
* Rewrote multijar to fix a race condition in the queue shared by the worker
threads. It now uses the multiprocessing Pool class and map() function.
* Fixed incorrect untar of Gblocks in build_3rd_party.sh