v0.5.3

What's Changed

BREAKING CHANGE
python 3.10 is no longer supported.

• BUILD Build with uv, hatchling, updated CI/pre-commit and SPEC 0 compliant package versions by @maltekuehl in #416
• MAINT fix quiet mode by @BorisMuzellec in #402
• DOC update .readthedocs.yaml to handle installation through uv by @BorisMuzellec in #421
• DOC remove stale development roadmap is was completed by @BorisMuzellec in #408

Full Changelog: v0.5.2...v0.5.3

v0.5.2

What's Changed

• MAINT move 1D variables from obsm and varmto obs and var by @BorisMuzellec in #394

Full Changelog: v0.5.1...v0.5.2

v0.5.1

What's Changed

• MAINT update actions/upload-artifacts to v4 by @BorisMuzellec in #364
• DOC update requirements rst by @BorisMuzellec in #373
• ENH DeseqStats n_cpus by @asistradition in #379

Full Changelog: v0.5.0...v0.5.1

v0.5.0

What's Changed

This version marks a shift towards formulaic designs, i.e. formula like "~condition + treatment", "~condition + condition:treatment", "~condition + exp(cofactor)", etc.

BREAKING CHANGE

python 3.9 is no longer supported.

DeseqDataSet

• Designs should now be provided to DeseqDataSet using the design argument, either in the form of a string representing a formulaic formula (e.g. "~condition + treatment", "~condition + condition:treatment", "~condition + exp(cofactor)"...), or an ndarray directly corresponding to a design matrix.
• design_factors is still supported but throws a DeprecationWarning
• continuous_factors is deprecated, as continuous type inference is handled by formulaic
• ref_level is deprecated
• Due to new decorated methods, DeseqDataSet is no longer picklable. A to_picklable_anndata() method was added to allow users to pickle results for later use.

DeseqStats

• Default contrasts are no longer supported, as they lead to too many errors
• Contrasts may be provided as before for categorical variables (e.g. ["treatment", "test", "control"]), or directly in the form of a contrast vector (a numpy array).
• For now, contrasts for continuous variables are directly specified with a contrast vector.
• lfc_shrink no longer supports a default coef argument

Changelog

• ENH Generic design support using formulaic by @BorisMuzellec in #328
• MAINT Deprecated pandas.DataFrameGroupBy.grouper by @BorisMuzellec in #332
• ENH add fit size factors type as an argument of DeseqDataSet by @regloeb in #345
• ENH Switch to formulaic-contrasts by @grst in #344
• BUG nbinomGLM returns before convergence by @BorisMuzellec in #349
• REF change type hints to PEP585 by @BorisMuzellec in #351
• ENH add access to control_genes as a DeseqDataSet attribute by @BorisMuzellec in #356

New Contributors

• @regloeb made their first contribution in #345
• @grst made their first contribution in #344

Full Changelog: v0.4.12...v0.5.0

v0.5.0pre2

What's Changed

• ENH add fit size factors type as an argument of DeseqDataSet by @regloeb in #345
• ENH Switch to formulaic-contrasts by @grst in #344

New Contributors

• @regloeb made their first contribution in #345
• @grst made their first contribution in #344

Full Changelog: v0.5.0pre1...v0.5.0pre2

v0.5.0pre1

What's Changed

This version marks a shift towards formulaic designs, i.e. formula like "~condition + treatment", "~condition + condition:treatment", "~condition + exp(cofactor)", etc.

BREAKING CHANGE

python 3.9 is no longer supported.

DeseqDataSet

• Designs should now be provided to DeseqDataSet using the design argument, either in the form of a string representing a formulaic formula (e.g. "~condition + treatment", "~condition + condition:treatment", "~condition + exp(cofactor)"...), or an ndarray directly corresponding to a design matrix.
• design_factors is still supported but throws a DeprecationWarning
• continuous_factors is deprecated, as continuous type inference is handled by formulaic
• ref_level is deprecated
• Due to new decorated methods, DeseqDataSet is no longer picklable. A to_picklable_anndata() method was added to allow users to pickle results for later use.

DeseqStats

• Default contrasts are no longer supported, as they lead to too many errors
• Contrasts may be provided as before for categorical variables (e.g. ["treatment", "test", "control"]), or directly in the form of a contrast vector (a numpy array).
• For now, contrasts for continuous variables are directly specified with a contrast vector.
• lfc_shrink no longer supports a default coef argument

Changelog

• ENH Generic design support using formulaic by @BorisMuzellec in #328
• MAINT Deprecated pandas.DataFrameGroupBy.grouper by @BorisMuzellec in #332

Full Changelog: v0.4.12...v0.5.0pre1

Acknowledgements

@jeandut worked on a first version of formulaic support in #181.
Thanks @grst and @umarteauowkin for your reviews and suggestions!

v0.4.12

What's Changed

• BUG set vst fit type to mean when parametric fails by @BorisMuzellec in #314
• ENH: Memory Optimizations & low_memory Flag by @asistradition in #324

Full Changelog: v0.4.11...v0.4.12

v0.4.11

This version aims to bring support for anndata 0.10.9.

What's Changed

• BUG set varm["replaced"] as a numpy array by @BorisMuzellec in #307
• MAINT Respect quiet=True in DeseqStats.lfc_shrink by @maltekuehl in #304

New Contributors

• @maltekuehl made their first contribution in #304

Full Changelog: v0.4.10...v0.4.11

v0.4.10

What's Changed

• BUG initialize beta_init when the design matrix is not full rank by @BorisMuzellec in #281
• MAINT Replace np.NaN with np.nan -> numpy 2.0 compat by @Zethson in #282
• BUG enable vst be used "blindly" and to fit its own dispersion by @laudmt in #268
• CHORE update code owners by @mandreux-owkin in #289
• MAINT summary function uses quiet to print by @yihming in #286
• ENH Poscount implementation & diag(XXT) optimization by @asistradition in #284
• BUG Inference n_cpus overwritten issue by @yihming in #293
• DOC specify what happens if n_cpus and inference are set by @umarteauowkin in #294
• DOC fix text format in DeseqStats docstring by @BorisMuzellec in #297
• ENH Remove count DataFrame from calculate_cooks by @asistradition in #292
• MAINT cast counts to int when initialising a DeseqDataSet from anndata by @BorisMuzellec in #299

New Contributors

• @Zethson made their first contribution in #282
• @laudmt made their first contribution in #268
• @yihming made their first contribution in #286
• @asistradition made their first contribution in #284
• @umarteauowkin made their first contribution in #294

Full Changelog: v0.4.9...v0.4.10

v0.4.9

What's Changed

• DOCS Fix latex display in sphinx doc by @tanguy-marchand in #266
• ENH Add adapt = False (prior_scale=1) option to lfc_shrink() by @awalsh17 in #267
• BUG update genewise dispersions and fitted dispersions of refitted genes by @BorisMuzellec in #274

New Contributors

• @tanguy-marchand made their first contribution in #266
• @awalsh17 made their first contribution in #267

Full Changelog: v0.4.8...v0.4.9