Panpipes

added

- added Tangram to `deconvolution_spatial`
- added scib metrics calculation to `integration`, using the [scib-metrics package](https://scib-metrics.readthedocs.io/en/latest/index.html)
- [extra documentation](https://panpipes-pipelines.readthedocs.io/en/latest/)

fixed

- fixed error in `vis`
- error occurred when only wanting to plot continuous or categorical variables (or neither), not both
- fixed error in `refmap`
- high threads was not recognised, now fixed.

dependencies
- All the dependencies have been updated.
- Python>=3.10 required
- added seeds to all scvi tasks for reproducibility

added

- added example multiome submission file 10X_h5
- added example multiome submission file cellranger
- workflows & tutorials for `qc_spatial`, `preprocess_spatial`, and `deconvolution_spatial` to readthedocs
- tutorial for `vis`
- added PCA parameters in pipeline_preprocess.py for PROT modality to fix issue 120
- added full control of dimred params for all modalities in pipeline_preprocess.py
- more info on custom genes format files added to documentation
- parsing summary files for cellranger multi version < 7
- added checks for n_pcs in run_neighbors_method_choice
- added filtering by HVF for atac

fixed

- changed typo in tutorial paths for clustering and deconvolution
- fix io to read cellranger outs folder for atac.
- fixes to refmap workflow
- typos & capitalization in the pipeline.yml files of `qc_spatial`, `preprocess_spatial`, and `deconvolution_spatial`, `vis`
- remove `assay`, `sample_prefix`, and `modalities` parameters from the `qc_spatial` pipeline.yml
- remove `sample_prefix` and `modalities` parameters from the `preprocess_spatial` pipeline.yml
- fixed error in `preprocess_spatial` when `filtering: run: False`
-> now able to run no filtering without needing to save the MuData in `filtered.data` before running the pipeline
- fixed error in `vis`
- change PARAMS['custom_markers_minimal'] -> PARAMS['custom_markers']['files']['minimal']
- fix to avoid rerunning HVF and explicitly check X layer before normalization in pipeline_preprocess.py
- fix plotting of umaps after batch correction
- fix fetching string scvi if present in mudata for wnn
- fixed lsi requirement for atac
- fixed top features for atac
- fixed filtering HVG for rna
- moved pynndescent to PyPi dependencies

dependencies

Big Change! the submission files for the `ingest` workflow have now changed! we require the paths to the Gene expression (RNA/GEX) and Protein (ADT) to have the following headers.

| sample_id | rna_path | rna_filetype | prot_path | prot_filetype |
| --------- | ----------- | ------------ | ------------ | ------------- |
| sampleX | path/to/rna | 10X_h5 | path/to/prot | 10x_h5 |
| | | | | |

See tutorials for examples of submission files.

added

- merged PR 111:
- LSI in panpipes_preprocess is run on the highly variable features
- n_comp for LSI

fixed

- changed all instances of ADT into PROT
- changed all instances of GEX to RNA
- changed the params to fix plotting as mentioned in issue 41
- typo in readme
- set default seaborn <=0.12.2 to avoid issue 104, 126

- set default matplotlib<=3.7.3 to avoid issue 104.

added

- Spatial data analysis is now included in panpipes
- panpipes qc_spatial
- panpipes preprocess_spatial
- panpipes_deconvolution_spatial

fixed

- make sure columns from individual modalities that are not in the multimodal outer obs can be used to

dependencies

- additional dependencies: squidpy, cell2location, openpyxl

- First public version of panpipes
- contains qc_mm, preprocess, intergration, clustering