Scirpy

Additions

- Add "TCRdist" as new metric ([502](https://github.com/scverse/scirpy/pull/502))

Fixes

- Fix issue with detecting the number of available CPUs on MacOD ([518](https://github.com/scverse/scirpy/pull/502))

Fixes

- Fix default value for `n_jobs` in `ir.tl.ir_query` that could lead to an error ([498](https://github.com/scverse/scirpy/pull/498)).
- Update description of D50 diversity metric in documentation ([499](https://github.com/scverse/scirpy/pull/498)).
- Fix `clonotype_modularity` not being able to store result in MuData in some cases ([504](https://github.com/scverse/scirpy/pull/504)).
- Fix issue with creating sparse matrices from generators with the latest scipy version ([504](https://github.com/scverse/scirpy/pull/504))

Backwards-incompatible changes

- Use the `umi_count` field instead of `duplicate_count` to store UMI counts. The field `umi_count` has been added to
the AIRR Rearrangement standard in [version 1.4](https://docs.airr-community.org/en/latest/news.html#version-1-4-1-august-27-2022) ([487](https://github.com/scverse/scirpy/pull/487)).
Use of `duplicate_count` for UMI counts is now discouraged. Scirpy will use `umi_count` in all `scirpy.io` functions.
It will _not_ change AIRR data that is read through `scirpy.io.read_airr` that still uses the `duplicate_count` column.
Scirpy remains compatible with datasets that still use `duplicate_count`. You can update your dataset using

python
adata.obsm["airr"]["umi_count"] = adata.obsm["airr"]["duplicate_count"]


Other

- the `io.to_dandelion` and `io.from_dandelion` interoperability functions now rely on the implementation provided by Dandelion itself ([483](https://github.com/scverse/scirpy/pull/483)).

Fixes

- Fix incompatibility with `scipy` 1.12 ([484](https://github.com/scverse/scirpy/pull/484))
- Fix incompatibility with `adjustText` 1.0 ([477](https://github.com/scverse/scirpy/pull/477))
- Reduce overall importtime by deferring the import of the `airr` package until it is actually used. ([473](https://github.com/scverse/scirpy/pull/473))

New features

- Speed up alignment distances by pre-filtering. There are two filtering strategies: A (lossless) length-based filter
and a heuristic based on the expected penalty per mismatch. This is implemented in the `FastAlignmentDistanceCalculator`
class which supersedes the `AlignmentDistanceCalculator` class, which is now deprecated. Using the `"alignment"` metric
in `pp.ir_dist` now uses the `FastAlignmentDistanceCalculator` with only the lenght-based filter activated.
Using the `"fastalignment"` activates the heuristic, which is significantly faster, but results in some false-negatives. ([456](https://github.com/scverse/scirpy/pull/456))
- Switch to [joblib/loky](https://joblib.readthedocs.io/en/latest/generated/joblib.Parallel.html) as a backend for parallel
processing in `pp.ir_dist`. Joblib enables to switch to alternative backends that support out-of-machine computing
(e.g. `dask`, `ray`) via the `parallel_config` context manager. Additionally, chunk sizes are now adjusted dynamically based on the problem size. ([473](https://github.com/scverse/scirpy/pull/473))

Documentation

- The default values of the distance calculator classes in `ir_dist.metrics` was unclear. The default value is now
set in the classes. In `pp.ir_dist` and `ir_dist.sequence_dist`, no cutoff argument is passed to the metrics
objects, unless one is explicitly specified (previously `None` was passed by default).

Breaking changes

- Reimplement `pp.index_chains` using numba and awkward array functions, achieving a significant speedup. This function
behaves exactly like the previous version _except_ that callback functions passed to the `filter` arguments
must now be vectorized over an awkward array, e.g. to check if a `junction_aa` field is present you could
previously pass `lambda x: x['junction_aa'] is not None`, now an accepted version would be
`lambda x: ~ak.is_none(x["junction_aa"], axis=-1)`. To learn more about native awkward array functions, please
refer to the [awkward array documentation](https://awkward-array.org/doc/main/reference/index.html). ([#444](https://github.com/scverse/scirpy/pull/444))

Additions

- The `clonal_expansion` function now supports a `breakpoints` argument for more flexible "expansion categories".
The `breakpoints` argument supersedes the `clip_at` parameter, which is now deprecated. ([439](https://github.com/scverse/scirpy/pull/439))

Fixes

- Fix that `define_clonotype_clusters` could not retreive `within_group` columns from MuData ([459](https://github.com/scverse/scirpy/pull/459))
- Fix that AIRR Rearrangment fields of integer types could not be written when their value was None ([465](https://github.com/scverse/scirpy/pull/465))

Fixes

- Fix that `clonotype_modularity` could not run with AnnData object ([421](https://github.com/scverse/scirpy/pull/421)).
- Fix usage of wrong column in 3k tutorial ([423](https://github.com/scverse/scirpy/pull/423))
- Change igraph dependency from `python-igraph` to `igraph` ([436](https://github.com/scverse/scirpy/pull/436))
- Fix that `group_abundance` didn't work when AIRR data was stored in a different MuData slot than `airr` ([438](https://github.com/scverse/scirpy/pull/438))