Changes
=======
0.4.5
-----
pyfasta split can handle > 52 files. (thanks Devtulya)

0.4.4
-----
fix pyfasta extract

0.4.3
-----
Add 0 or 1-based intervals in sequence() thanks @jamescasbon

0.4.2
-----
update for latest numpy (can't close memmap)

0.4.1
-----
check for duplicate headers.

0.4.0
-----
* add key_fn kwarg to constuctor

0.3.9
-----
* only require 'r' (not r+) for memory map.

0.3.8
-----
* clean up logic for mixing inplace/non-inplace flattened files.
  if the inplace is available, it is always used. 

0.3.6/7
-------
* dont re-flatten the file every time!
* allow spaces before and after the header in the orginal fasta.

0.3.5
-----

* update docs in README.txt for new CLI stuff.
* allow flattening inplace.
* get rid of memmap (results in faster parsing).

0.3.4
-----

* restore python2.5 compatiblity.
* CLI: add ability to exclude sequence from extract
* CLI: allow spliting based on header.

0.3.3
-----

* include this file in the tar ball (thanks wen h.)

0.3.2
-----

* separate out backends into records.py

* use nosetests (python setup.py nosetests)

* add a TCRecord backend for next-gen sequencing availabe if tc is (easy-)installed.

* improve test coverage.