C API 0.99.13

Bugfix release

Fixes

• Fix segfault when very large columns overflow
  (@bhaller, @benjeffery, #1509, #1511).

Python 0.3.6

Minor feature release

Notebook detailing the new features here: https://gist.github.com/benjeffery/aff619fd8da6799bccab81655c391965

Breaking changes

• Mutation.position and Mutation.index which were deprecated in 0.2.2 (Sep '19) have
  been removed.

Features

• Add direct, copy-free access to the arrays representing the quintuply-linked structure
  of Tree (e.g. left_child_array). Allows performant algorithms over the tree
  structure using, for example, numba
  (@jeromekelleher, #1299, #1320).

• Add fancy indexing to tables. E.g. table[6:86] returns a new table with the
  specified rows. Supports slices, index arrays and boolean masks
  (@benjeffery, #1221, #1348, #1342).

• Add Table.append method for adding rows from classes such as SiteTableRow and
  Site (@benjeffery, #1111, #1254).

• SVG visualization of a tree sequence can be restricted to displaying between left
  and right genomic coordinates using the x_lim parameter. The default settings
  now mean that if the left or right flanks of a tree sequence are entirely empty,
  these regions will not be plotted in the SVG (@hyanwong, #1288).

• SVG visualization of a single tree allows all mutations on an edge to be plotted
  via the all_edge_mutations param (@hyanwong,#1253, #1258).

• Entity classes such as Mutation, Node are now python dataclasses
  (@benjeffery, #1261).

• Metadata decoding for table row access is now lazy (@benjeffery, #1261).

• Add html notebook representation for Tree and change Tree.__str__ from dict
  representation to info table. (@benjeffery, #1269, #1304).

• Improve display of tables when printed, limiting lines set via
  tskit.set_print_options (@benjeffery,#1270, #1300).

• Add Table.assert_equals and TableCollection.assert_equals which give an exact
  report of any differences. (@benjeffery,#1076, #1328)

Changes

• In drawing methods max_tree_height and tree_height_scale have been deprecated
  in favour of max_time and time_scale
  (@benjeffery,#1262, #1331).

Fixes

• Tree sequences were not properly init'd after unpickling
  (@benjeffery, #1297, #1298)

C API 0.99.12

Minor feature release

Breaking changes

• Removed TSK_NO_BUILD_INDEXES.
  Not building indexes is now the default behaviour of tsk_table_collection_dump and related functions.
  (@molpopgen, #1327, #1337).

Features

• Add tsk_*_table_extend methods to append to a table from another
  (@benjeffery, #1271, #1287)

Python 0.3.5

Breaking changes

• tskit now requires Python 3.7 (@benjeffery, #1235)

Features

• SVG visualization plots mutations at the correct time, if it exists, and a y-axis,
  with label can be drawn. Both x- and y-axes can be plotted on trees as well as
  tree sequences (@hyanwong,#840, #580, #1236)

• SVG visualization now uses squares for sample nodes and red crosses for mutations,
  with the site/mutation positions marked on the x-axis. Additionally, an x-axis
  label can be set (@hyanwong,#1155, #1194, #1182, #1213)

• Add parents column to the individual table to allow recording of pedigrees
  (@ivan-krukov, @benjeffery, #852, #1125, #866, #1153, #1177, #1192 #1199).

• Added Tree.generate_random_binary static method to create random
  binary trees (@hyanwong, @jeromekelleher, #1037).

• Change the default behaviour of Tree.split_polytomies to generate
  the shortest possible branch lengths instead of a fixed epsilon of
  1e-10. (@jeromekelleher, #1089, #1090)

• Default value metadata in add_row functions is now schema-dependant, so that
  metadata={} is no longer needed as an argument when a schema is present
  (@benjeffery, #1084).

• default in metadata schemas is used to fill in missing values when encoding for
  the struct codec. (@benjeffery, #1073, #1116).

• Added canonical option to table collection sorting (@mufernando,
  @petrelharp, #705)

• Added various arguments to TreeSequence.subset, to allow for stable
  population indexing and lossless node reordering with subset.
  (@petrelharp, #1097)

Changes

• Allow mutations that have the same derived state as their parent mutation.
  (@benjeffery, #1180, #1233)

• File minor version change to support individual parents

C API 0.99.11

Features

• Add parents to the individual table to enable recording of pedigrees
  (@ivan-krukov, @benjeffery, #852, #1125, #866, #1153, #1177, #1199).

• Added a tsk_table_collection_canonicalse method, that allows checking for equality between
  tables that are equivalent up to reordering (@petrelharp, @mufernando, #1108).

• Removed a previous requirement on tsk_table_collection_union, allowing for unioning of
  new information both above and below shared history (@petrelharp, @mufernando, #1108).

• Support migrations in tsk_table_collection_sort. (@jeromekelleher,
  #22, #117, #1131).

Breaking changes

• Method tsk_individual_table_add_row has an extra arguments parents and parents_length.

• Add an options argument to tsk_table_collection_subset (@petrelharp, #1108),
  to allow for retaining the order of populations.

• Mutation error codes have changed

Changes

• Allow mutations that have the same derived state as their parent mutation.
  (@benjeffery, #1180, #1233)

• File minor version change to support individual parents

C API 0.99.10

Minor change to internal APIs

C API 0.99.9

Minor Feature Release

Features

• Add TSK_KEEP_UNARY_IN_INDIVIDUALS flag to simplify, which allows the user to
  keep unary nodes only if they belong to a tabled individual. This is useful for
  simplification in forwards simulations (@hyanwong, #1113, #1119).

Python 0.3.4

Minor bugfix release.

Bugfixes

• Reinstate the unused zlib_compression option to tskit.dump, as msprime < 1.0 still uses it (@jeromekelleher, #1067).

Python 0.3.3

Minor feature release

Features

• Add TreeSequence.genetic_relatedness for calculating genetic relatedness between
  pairs of sets of nodes (@brieuclehmann, #1021, #1023, #974, #973, #898).

• Expose TreeSequence.coiterate() method to allow iteration over 2 sequences
  simultaneously, aiding comparison of trees from two sequences
  (@jeromekelleher, @hyanwong, #1021, #1022).

• tskit is now supported on, and has wheels for, python3.9
  (@benjeffery, #982, #907).

• Tree.newick() now has extra option include_branch_lengths to allow branch
  lengths to be omitted (@hyanwong, #931).

• Added Tree.generate_star static method to create star-topologies (@hyanwong,
  #934).

• Added Tree.generate_comb and Tree.generate_balanced methods to create
  example trees. (@jeromekelleher, #1026).

• Added equals method to TreeSequence, TableCollection and each of the tables which
  provides more flexible equality comparisons, for example, allowing
  users to ignore metadata or provenance in the comparison
  (@mufernando, @jeromekelleher, #896, #897,
  #913, #917).

• Added __eq__ to TreeSequence
  (@benjeffery, #1011, #1020).

• ts.dump and tskit.load now support reading and writing file objects such as
  FIFOs and sockets (@benjeffery, #657, #909).

• Added tskit.write_ms for writing to MS format
  (@saurabhbelsare, #727, #854).

• Added TableCollection.indexes for access to the edge insertion/removal order indexes
  (@benjeffery, #4, #916).

• The dictionary representation of a TableCollection now contains its index
  (@benjeffery, #870, #921).

• Added TreeSequence._repr_html_ for use in jupyter notebooks
  (@benjeffery, #872, #923).

• Added TreeSequence.__str__ to display a summary for terminal usage
  (@benjeffery, #938, #985).

• Added TableCollection.dump and TableCollection.load. This allows table
  collections that are not valid tree sequences to be manipulated
  (@benjeffery, #14, #986).

• Added nbytes method to tables, TableCollection and TreeSequence which
  reports the size in bytes of those objects
  (@jeromekelleher, @benjeffery, #54, #871).

• Added TableCollection.clear to clear data table rows and optionally
  provenances, table schemas and tree-sequence level metadata and schema
  (@benjeffery, #929, #1001).

Bugfixes

• LightWeightTableCollection.asdict and TableCollection.asdict now return copies
  of arrays (@benjeffery, #1025, #1029).

• The map_mutations method previously used the Fitch parsimony method, but this
  does not produce parsimonious results on non-binary trees. We now now use the
  Hartigan parsimony algorithm, which does (@jeromekelleher,
  #987, #1030).

• The flag argument to tables' add_row was treating the value as signed
  (@benjeffery, #1027, #1031).

Breaking changes

• The argument to ts.dump and tskit.load has been renamed file from path.
• All arguments to Tree.newick() except precision are now keyword-only.
• Renamed ts.trait_regression to ts.trait_linear_model.

C API 0.99.8

Minor feature release

New features

• Add tsk_treeseq_genetic_relatedness for calculating genetic relatedness between
  pairs of sets of nodes (@brieuclehmann, #1021, #1023, #974, #973, #898).

• Exposed tsk_table_collection_set_indexes to the API
  (@benjeffery, #870, #921).

Breaking changes

• Added an options argument to tsk_table_collection_equals
  and table equality methods to allow for more flexible equality criteria
  (e.g., ignore top-level metadata and schema or provenance tables).
  Existing code should add an extra final parameter 0 to retain the
  current behaviour (@mufernando, @jeromekelleher,
  #896, #897, #913, #917).

• Changed default behaviour of tsk_table_collection_clear to not clear
  provenances and added options argument to optionally clear provenances
  and schemas (@benjeffery, #929, #1001).

• Renamed tsk_treeseq_trait_regression to tsk_treeseq_trait_linear_model.