Added
- New `--no-reformat` option for `interleave-reads.py` script disables default
read name correction behavior.
- New `HashSet` data structure for managing collections of k-mer hashes and
tags.
- khmer package version now included in `.info` files.
- New `-o|--outfile` option for `filter-abund-single.py` script.
- New sandbox script `extract-compact-dbg.py` for computing a compact de Bruijn
graph from sequence data.
- New `--quiet` flag to several scripts, silencing diagnostic messages in
terminal output.
- Support for human-friendly memory requests (2G instead of 2000000000 or 2e9).
- Support for variable-coverage trimming in the `filter-abund-single.py` script.
- Several simple examples of the Python API and the C++ API in
`examples/python-api` and `examples/c++-api`, respectively.
- New `assemble_linear_path` function for baiting unambiguous contigs with a
seed k-mer from a hashtable.
- Support for assembling directly from k-mer graphs, and a new
JunctionCountAssembler class.
- Add --info flag for obtaining citation information.
- Added Counttable and Nodetable to support non-reversible hashing
functionality and k > 32.
- Add a new storage class using half a byte per entry. Exposed as
SmallCounttable and SmallCountgraph.
- Added `cleaned_seq` attribute to `khmer.Read` class which provides a cleaned
version of the sequence of each read.
- Added --summary-info to trim-low-abund.py to record run information in a file.
- `Nodetable`, `Counttable` and `SmallCounttable` use murmur hash 3 as hash
function. This means they support kmers longer than 32 bases but means
the hashes are not reversible.
Changed
- Suppress display of -x and -N command line options in script help messages.
- Switch from nose to py.test as the testing framework.
- Switch from internally managed Jenkins setup to Travis CI for continuous
integration testing.
- Renamed core data structures: CountingHash --> Countgraph,
Hashbits --> Nodegraph.
- Replaced the IParser and FastxParser classes with a single ReadParser class.
Different input formats are supported by templating ReadParser with a reader
class.
- Renamed `consume_fasta` and related functions to `consume_seqfile`, with
support for reading sequences from additional formats pending.
- Changed Sphinx documentation theme to "guzzle".
Fixed
- Bug in compressed(gzip) streaming output from scripts
- The hashbits `update_from` function to correctly track occupied bins for
calculating FPR.
- Bug in the `filter-abund.py` script when `--gzip` and `-o` flags are used
simultaneously.
- Bug in the hashtable `get_kmers` function based on incorrect usage of the
`substr` function.
- Bug in `broken_paired_reader` related to dropping short reads when
`require_paired` is set.
- Bug related to handling lowercase [acgtn] characters in input data.
- Bug in `load-graph.py` that calculated required graph space incorrectly.
- Fix loading of empty partion map files