v2.1.6

Bugfixes

• Fixed op.translate (and every op.resource call for non-human organisms) failing with HTTP 404 after HGNC dropped the genenames/ subtree from the EBI FTP mirror; the HCOP fifteen-column files are now fetched from the new HGNC Google Cloud Storage bucket (#303)
• Removed a redundant pd.read_csv call in op.translate that re-downloaded the HCOP file a second time and bypassed the retry logic in _download
• Fixed bm.benchmark ground-truth matrix row order not matching score matrix after pivot, which silently misaligned perturbation labels when obs names sort differently from their original order (#302)

v2.1.5

Added

• pl.volcano now accepts a gene name (str) or list of gene names (list[str]) for the top parameter to annotate specific features on volcano plots

Changes

• Refactored ds.ensmbl_to_symbol to reuse _download and fixed mirror fallback to actually switch between Ensembl mirrors
• Made tqdm progress bar compact in _download

Bugfixes

• Fixed pl.dotplot ignoring vcenter=0 due to falsy check (#293)
• Fixed _log setting the root logging level to INFO for all packages (#296)

v2.1.4

2.1.4

Changes

• Added alternative argument to decoupler.mt.query_set. By default 'greater', before it was 'two-sided'
• Unpinned scipy version limit

Bugfixes

• Fixed missing progressbar for decoupler._download._download
• Added missing decoupler.mt.query_set documentation

v2.1.3

• pp.adjmat now returns the same features as used as input instead of the subset of net
• pp.pseudobulk now returns the same order features as used as input instead of shuffling them
• Added a dedicated header and 5 attempts to _download to mitigate 429 Client Error from Zenodo downloads

v2.1.2

Added

• pp.query_set to test overlap between a given feature set against a database of sets

Changes

• tl.rankby_obsm now accepts AnnData.obs column names specified in the obs_keys argument
• Most plotting functions now accept extra arguments through kw_ arguments.
• p-values now are corrected using a custom numba-optimized version of scipy.stats.false_discovery_control called _fdr_bh_axis1_numba

v2.1.1

2.1.1

Added

• mypy checks in CI
• notebook checks in CI

Bugfixes

• Fixed error in decoupler.pp.pseudobulk when adata.obs_names were not unique, now throws verbose error
• Fixed corner case in decoupler.mt.gsea when p-values were infitite and could not be corrected

v2.1.0

Changes

• Updated logo
• decoupler._download._download now returns bytes instead of a dataframe. To transform to pandas.DataFrame use decoupler._download._bytes_to_pandas
• Enrichment methods and pseudobulking now work with backed AnnData objects, useful when working with big datasets and memory is limited

Bugfixes

• Fixed error in pl.obsm where default value of cmap_obs was not properly set.

v2.0.7

Added pre-commit

v2.0.6

Modified links and paths to follow scverse repo

v2.0.5

Fixed error message when extra dependencies where not installed.