Pytximport

API
- Added experimental support for SummarizedExperiment as `output_type`/`output_format`
- Fixed an RSEM bug where a minimum length requirement of 1 was not being enforced

API
- Added support for .csv transcript to gene maps
- Added experimental inferential replicate support for oarfish and psicem quantification files
- Moved `ignore_bar` and `ignore_transcript_version` logic from `utils.convert_transcripts_to_genes` to `core.tximport`

Other changes
- General performance improvements
- Vectorized get_median_length_over_isoform
- Added support for pyarrow for reading in .tsv/.csv files
- Removed unnecessary copy operations
- Added support for .csv transcript to gene maps
- Added test for RSEM and Piscem files
- Updated the documentation to include recommendations for installation options and added code examples to docstrings

API
- Allows the source column and the target column from the GTF to be explicitly set when creating a transcript-to-gene or transcript-name-to-transcript-mapping
- Removes unused `sparse` and `sparse_threshold` arguments

Other changes
- Adds regular cron job CI testing to detect changed dependencies

API
- Added support for RSEM gene-level quantification files. Previously, it was already possible to read RSEM transcript level data from a .tsv file. Now, not only can RMEM transcript data be read in by setting `data_type` to `rsem`, but if `gene_level` is set to True, existing gene-level data can be read in and aggregated by `pytximport`.
- Added support for gzipped files when reading in tabular data with `data_type` set `tsv`.

Other changes
- `pytximport` is now available through Bioconda and an install badge has been added to the README and the documentation.
- Added unit tests to check RSEM gene-level quantification for correctness.

API
- Added utility functions to create transcript-to-gene maps from Biomart or .gtf files.
- Added a flag to allow overwriting existing files.

Other changes
- Added unit tests for file saving via programatic (non-CLI) usage.

API
- Added support for inferential replicates via the `inferential_replicates`, `inferential_replicate_variance` and `inferential_replicate_transformer` arguments with support for salmon and kallisto.
- Fixed a bug where custom column names were not applied.
- Fixed a bug when saving .csv files from AnnData objects.

Other changes
- Added unit tests for file saving via programatic (non-CLI) usage.