sbwt 0.6.2

Indexing sets of DNA k-mers with the spectral Burrow-Wheeler transform.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139

# Changelog

## v0.6.2

Removed some unnecessary implementation details from the docs.

## v0.6.1

The BuildByLibsais construction algorithm now runs fully in linear time at the cost of using O(n log n + n log k) extra bits. BuildByBoundedSuffixSort uses the same functions, so it also pays the same memory cost.

## v0.6.0

### New features

* A variable-order DBG interface, with optional k-mer counts.
* Two new construction algorithms: one based on bounded-context suffix sorting, another based on suffix sorting via libsais (behind the feature gate "libsais"). The builder API has been reworked to accommodate these, and it is **not backwards compatible**.

### Bugfixes

* Fixed a bug in the set difference operation which lead to a corrupt index.
* Fixed a bug where redundant dummy nodes were sometimes left behind by set operations.
* Fixed a crash when running set operations on very small inputs.

### Performance

* Disk-based construction: better batching, and introduced parallelism for later stages of the construction pipeline.

## v0.5.0

- Added support for set operations (union, intersection, difference).
  - API: new `set_operations` module (`src/set_operations`) exposing the
    core functions `merge`, `intersect` and `difference`, together with
    helper types such as `MergeInterleaving`.
  - Union is based on the existing merge code. It now also cleans up redundant dummy nodes.
- Fixed bug in `push_labels_forward_compact` which occurred in some cases when some nucleotide does not occur in the SBWT at all.

## v0.4.3

* Update to jseqio v0.1.7. This required adding the Sync trait bound to `run_from_fasta` and `run_from_fastq` in the builder.

## v0.4.2

### Performance

* Significantly speed up `push_labels_forward`, speeding up e.g. the SBWT merge algorithm.
* Better parallelism in unitig iteration and exporting.

### API features

* Expose a function for a single matching statistics update step.
* Add a matching statistics iterator with a length bound below k.
* Make functions related to unitigs public.
* Make the dummy marking algorithm public and move it to SbwtIndex.
* Provide access to DBG dummy marks.
* Eliminate most dependencies on `simple_sds_sbwt` from the public API. Some dependencies remain at `streaming_index.rs` and `sdsl_compatibility.rs`.

### Bugfixes

* Make the DBG implementation work even if there are redundant dummies.
* Fix empty ranges in parallel segmentation.
* Fix past-the-end bits in bitvec.

## v0.4.1

New algorithms and performance improvements. This is a major update, but still backward compatible.

* Add an algorithm to merge two SBWTs.
* Improvements in the parallelism and performance of the in-memory construction algorithm.
* In-memory construction is no longer behind a feature gate.
* Rename BitPackedKmerSorting to BitPackedKmerSortingDisk. The old type name works as an alias but is marked as deprecated.
* Add a streaming matching statistics iterator.

## v0.4.0

- Added `iter_unitigs_with_callback`. This can break existing code because now `parallel_export_unitigs` also takes the
  number of threads as an input parameter, whereas previously it used all available threads.

## v0.3.11

- Fix SbwtIndex documentation not showing up.

## v0.3.10

- Fix `push_labels_forward` behavior at root: now fills in dollars, as documentation says.

## v0.3.9

- Make `build_last_column` public.

## v0.3.8

- Add disk usage reporting during construction.
- Make `push_labels_forward` public.

## v0.3.7

- Parallelise the subroutine `push_labels_forward`. This speeds up the k-mer dump, LCS array construction and DBG construction.
- Add a CLI command `build-lcs` to build the LCS array from the SBWT on the command line.
- Make CLI commands `lookup`, `matching-statistics`, `dump-kmers` and `dump-unitigs` all print to stdout if an output file is not given.

## v0.3.6

- Fixed a corner case when importing an index with an empty precalc table from the C++ format. This caused a crash at query time, which is fixed in this patch.


## v0.3.5

- Add `load_from_cpp_plain_matrix_format` to load the index from the C++ API
  or CLI format: https://github.com/algbio/SBWT.

## v0.3.4

- Add an optional in-memory construction implementation by Tommi Mäklin: https://github.com/jnalanko/sbwt-rs-cli/pull/4. This can be built with by enabling the feature `bpks-mem`.
- Fixes to make the crate work with WebAssembly.
- Parallel C-array construction.

## v0.3.3

- Documentation: Add an example using Needletail for FASTX parsing.
- Change usize to u32 in `IS_DNA` for platform-independence.

## v0.3.2

- Fixes to documentation.

## v0.3.1

- Added function `from_subset_seq` to SbwtIndex.

## v0.3.0

- Added serialization magic string to the start of the serialization format. This breaks compatibility with files generated using the previous versions.
- Made some functions generic over the particular implementation of SubsetSeq in preparation for the future when we will have multiple implementations.
- Added to the documentation a link to the Github repository of the CLI.

## v0.2.1

- Added CHANGELOG.md
- Updated jseqio to version 0.1.3. This makes it so that lower-case characters are made upper-case during parsing. Previously k-mers containing lower-case characters were ignored.