Biom-format

----------

Minor fixes, released on January 29, 2014

Bug fixes:

* Improve error message when trying to load an HDF5 file without h5py being
installed.
* Allow validating json files when h5py is not installed.

----------

Minor fixes, released on December 18, 2014

Bug fixes:

* Remove syntax error from `normalize_table.py`.
* `Table.to_json` was not serializing empty tables properly, see 571
* `biom summarize-table` could not handle empty tables, see 571

----------

Minor fixes and performance improvements, released on November 19th 2014

Changes:

* The collapsing function to `Table.collapse` is now passed the entire table to
allow for more complex collapses (e.g., median, random selection, etc). See
544, 545 and 547.
* Updated version strings in the project to be
[Semantic Versioning](www.semver.org)-stlye. This better matches with other
open source python projects, and plays nicer with pip.
* Conversion from TSV now takes less memory. See 551.
* Parameter header_mark has been removed from _extract_data_from_tsv()
in table.py
* Order of magnitude improvement in parsing HDF5 BIOM tables, see 529
* Added `Table.length`, see 548
* Order of magnitude performance increase in `Table.nonzero`, see 538

Bug fixes:

* Ensure that a copy is performed in `Table.subsample`
* Avoided a memory leak when checking if a table is JSON or TSV, see 552.

--------

Format finalization, released on August 7th 2014

New features:

* Group metadata (e.g., a phylogenetic tree) can now be stored within the HDF5
representation. These data are available within the `Table` object
* Matrix data can now be accessed by the ``Table.matrix_data`` property
* ``Table`` IDs are now accessed via the ``Table.ids`` method
* ``Table`` metadata are now accessed via the ``Table.metadata`` method
* New method ``Table.update_ids``, which allows for updating the ids along
either axis.
* added ``normalize-table`` option to optparse and HTML interfaces which
utilizes the new TableNormalizer command from ``table_normalizer.py``

Changes:

* Metadata are now stored in individual datasets within HDF5. This resulted in
a change to the BIOM-Format spec which has now been bumped to format
version 2.1.
* ``Table.collapse`` ``min_group_size`` is now 1 by default, see 480
* General improvements to BIOM 2.x online documentation
* ``Table.pa`` now supports negative values
* dropped old, unused scripts
* added ``Table.iter_pairwise``
* added ``Table.min`` and ``Table.max``, see 459
* iter methods now support dense/sparse
* added ``Table.matrix_data`` property
* ``Table.filter`` yields a sparse vector, see 470
* ``Table.subsample`` can now sample by IDs (e.g., get a random subset of
samples or observations from a ``Table``).
* ``biom.util.generate_subsamples`` will generate an infinite number of
subsamples and can be used for rarefaction.
* ``biom summarize-table`` can now operate on observations.
* 10% performance boost in ``Table.subsample``, see 532

Bug fixes:

* ``Table.transform`` operates on full vectors now, see 476
* ``biom convert`` now handles taxonomy strings correctly, see 504
* ``Table.sort_order`` was not retaining ``Table.type``, see 474
* ``convert_biom_to_table`` now uses ``load_table``, see 478
* ``Table.pa`` now handles negative values, see 492
* ``Table.copy`` was not retaining ``Table.type``, see 494

----------

Bug fix release, released on June 3rd 2014

Changes:

* Light weight loading mechanism (`biom.load_table`) added
* `Table.data` now has a default axis
* Convert documentation updated
* Quick start page added to documentation

Bug fixes:

* missing fields from JSON representation reintroduced
* `TableConverter` works as expected

----------

Major release, released on May 15th 2014

Changes:

* NumPy 1.7 or above is required
* Support for HDF5
* Codebase is PEP-8 compliant
* CSMat has been removed and Scipy is now a required dependency
* Requires pyqi 0.3.2
* New HTML interface
* No longer dependent on dateutil
* `Table.bin_samples_by_metadata` and `Table.bin_observations_by_metadata` have
been combined into `Table.partition`, which takes an axis argument
* `Table.collapse_samples_by_metadata` and
`Table.collapse_observations_by_metadata` have been combined into
`Table.collapse`, which now takes an axis argument
* `Table.filter_samples` and `Table.filter_observations` have been combined
into `Table.filter`, which now takes an axis argument
* `Table.transform_samples` and `Table.transform_observations` have been
combined into `Table.transform`, which now takes an axis argument
* `Table.norm_sample_by_observation` and `Table.norm_observation_by_sample`
have been combined into `Table.norm`, which now takes an axis argument
* `Table.iter_samples` and `Table.iter_observations` have been combined into
`Table.iter`, which now takes an axis argument
* `Table.iter_sample_data` and `Table.iter_observation_data` have been combined
into `Table.iter_data`, which now takes an axis argument
* `Table.get_sample_index` and `Table.get_observation_index` have been combined
into `Table.get_index`, which now takes an axis argument
* `Table.add_sample_metadata` and `Table.add_observation_metadata` have been
combined into `Table.add_metadata`, which now takes an axis argument
* `Table.sample_data` and `Table.observation_data` have been combined into
`Table.data`, which now takes an axis argument
* `Table.sample_exists` and `Table.observation_exists` have been combined into
`Table.exists`, which now takes an axis argument
* `Table.sort_by_sample_ids` and `Table.sort_by_observation_ids` have been
combined into `Table.sort`, which now takes an axis argument
* `Table.sort_sample_order` and `Table.sort_observation_order` have been
combined into `Table.sort_order`, which now takes an axis argument
* `Table.norm_samples_by_metadata` and `Table.norm_observations_by_metadata`
have been removed
* Added `Table.metadata` to allow fetching of metadata by an ID instead of just
by index
* Added `Table.pa` for conversion to presence/absence
* Added `Table.subsample` for randomly subsampling data
* `Table` now embraces numpydoc