Cutadapt

-----------------

* At least 25% faster processing of .csfasta/.qual files due to faster parser.
* Between 10% and 30% faster writing of gzip-compressed output files.
* Support 5' adapters in colorspace, even when no primer trimming is requested.
* Add the ``--info-file`` option, which has a line for each found adapter.
* Named adapters are possible. Usage: ``-a My_Adapter=ACCGTA`` assigns the name "My_adapter".
* Improve alignment algorithm for better poly-A trimming when there are sequencing errors.
Previously, not the longest possible poly-A tail would be trimmed.
* James Casbon contributed the ``--discard-untrimmed`` option.

-----------------

* Allow to "anchor" 5' adapters (``-g``), forcing them to be a prefix of the read.
To use this, add the special character ``^`` to the beginning of the adapter sequence.
* Add the "-N" option, which allows 'N' characters within adapters to match literally.
* Speedup of approx. 25% when reading from .gz files and using Python 2.7.
* Allow to only trim qualities when no adapter is given on the command-line.
* Add a patch by James Casbon: include read names (ids) in rest file
* Use nosetest for testing. To run, install nose and run "nosetests".
* When using cutadapt without installing it, you now need to run ``bin/cutadapt`` due to
a new directory layout.
* Allow to give a colorspace adapter in basespace (gets automatically converted).
* Allow to search for 5' adapters (those specified with ``-g``) in colorspace.
* Speed up the alignment by a factor of at least 3 by using Ukkonen's algorithm.
The total runtime decreases by about 30% in the tested cases.
* allow to deal with colorspace FASTQ files from the SRA that contain a fake
additional quality in the beginning (use ``--format sra-fastq``)

-----------------

* ASCII-encoded quality values were assumed to be encoded as ascii(quality+33).
With the new parameter ``--quality-base``, this can be changed to ascii(quality+64),
as used in some versions of the Illumina pipeline. (Fixes issue 7.)
* Allow to specify that adapters were ligated to the 5' end of reads. This change
is based on a patch contributed by James Casbon.
* Due to cutadapt being published in EMBnet.journal, I found it appropriate
to call this release version 1.0. Please see
http://journal.embnet.org/index.php/embnetjournal/article/view/200 for the
article and I would be glad if you cite it.
* Add Galaxy support, contributed by Lance Parsons.
* Patch by James Casbon: Allow N wildcards in read or adapter or both.
Wildcard matching of 'N's in the adapter is always done. If 'N's within reads
should also match without counting as error, this needs to be explicitly
requested via ``--match-read-wildcards``.

-------------------

* Fix issue 20: Make the report go to standard output when ``-o``/``--output`` is
specified.
* Recognize `.fq` as an extension for FASTQ files
* many more unit tests
* The alignment algorithm has changed. It will now find some adapters that
previously were missed. Note that this will produce different output than
older cutadapt versions!

Before this change, finding an adapter would work as follows:

- Find an alignment between adapter and read -- longer alignments are
better.
- If the number of errors in the alignment (divided by length) is above the
maximum error rate, report the adapter as not being found.

Sometimes, the long alignment that is found had too many errors, but a
shorter alignment would not. The adapter was then incorrectly seen as "not
found". The new alignment algorithm checks the error rate while aligning and only
reports alignments that do not have too many errors.

-------------------

* now compatible with Python 3
* Add the ``--zero-cap`` option, which changes negative quality values to zero.
This is a workaround to avoid segmentation faults in BWA. The option is now
enabled by default when ``--bwa``/``--maq`` is used.
* Lots of unit tests added. Run them with ``cd tests && ./tests.sh``.
* Fix issue 16: ``--discard-trimmed`` did not work.
* Allow to override auto-detection of input file format with the new ``-f``/``--format``
parameter. This mostly fixes issue 12.
* Don't break when input file is empty.

-------------------

* Install a single ``cutadapt`` Python package instead of multiple Python
modules. This avoids cluttering the global namespace and should lead to less
problems with other Python modules. Thanks to Steve Lianoglou for
pointing this out to me!
* ignore case (ACGT vs acgt) when comparing the adapter with the read sequence
* .FASTA/.QUAL files (not necessarily colorspace) can now be read (some
454 software uses this format)
* Move some functions into their own modules
* lots of refactoring: replace the fasta module with a much nicer seqio module.
* allow to input FASTA/FASTQ on standard input (also FASTA/FASTQ is
autodetected)