Skip to content

CHANGELOG.md

Latest commit

 

History

History
347 lines (212 loc) · 9.67 KB

CHANGELOG.md

File metadata and controls

347 lines (212 loc) · 9.67 KB
Raw

Changelog

[1.12.0] - 2025-02-14

Changed

  • Replace python-glmnet with my own that has an updated compile setup and allows for numpy > 1.26
  • Use numpy >= 2.0
  • the main PLIER function now has options to enable debug logging and to save the log to a file

Fixed

  • Logging should work better
  • Random scattered fixes for numpy >= 2

[1.11.1] - 2024-11-XX

Changed

  • Replaced numpy.ndarray where it was used for typing with numpy.testing.NDArray

[1.11.1] - 2024-11-04

Fixed

  • Caught a few stray variable names that had not had their capitalizaition changed
  • Removed a decorator from PLIER that forced passing a value when a default was there

[1.10.0] - 2024-11-01

Changed

  • Renamed a lot of functions and variables to follow standard Python naming conventions (i.e make ruff stop complaining)
  • Reworked a lot of the unit tests

1.8.0 - 2022-09-23

Changed

  • Changed svdres argument to PLIER from an np.ndarray to a Dict[str, np.ndarray] to reflect what is actually returned by scipy.linalg.svd
  • Added arguments to all functions that use tqdm to either remove the output from tqdm-based progressbars after completetion or just to disable them completely.

Fixed

  • Added a implementation to num_pc.compute_uu to handle pandas dataframes (e.g. convert to a ndarray and use the its implementation)
  • If a pre-computed svd is passed to PLIER, it now checks that the number of columns in the svd['v'] member matches the number of columns in the data, not that the svd['v'] shape tuple matches

1.7.0 - 2022-09-21

Changed

  • replaced pyplier.utils.rowNorm with pyplier.utils.zscore as that is what it was anyway
    • also, it is no self applies - you will need to run it over rows manually (i.e. df.apply(zscore, axis=1))
  • removed pyplier.utils.colNorm since it was unused

Fixed

  • pyplier.utils.zscore should be able to handle rows where all the values are the same (return 0, since all values are right on the standard deviation of 0)

1.6.0 - 2022-09-21

Fixed

  • fixed an issue when reading a PLIERResult object stored as an h5 has empty lists as part of the "heldOutGenes" member

1.5.5 - 2022-09-13

Fixed

  • saving non-integer numeric dictionaries in PLIERResults.withPrior to hdf5 no longer results in their being inappropriately cast to integers

1.5.4 - 2022-09-13

Added

  • Automated unit testing on Github

Changed

  • Instead of using the stale civisanalytics/python-glmnet, switch to the slightly-less stale replicahq/python-glmnet (which uses pep518-compliant install, obviating any issues with needing numpy installed prior to glmnet)

1.5.3 - 2022-09-12

Fixed

  • Another attempt at fixing an issue where a PLIERResults object could not be saved as hdf5 because of the "LV index" text column in the summary dataframe (and this is why you should write unit tests prior to saying you "fixed" something)

1.5.2 - 2022-09-12

Fixed

  • Fixed an issue where a PLIERResults object could not be saved as hdf5 because of the "LV index" text column in the summary dataframe

1.5.1 - 2022-09-12

Fixed

  • Removed stray glmnet-py dep

1.5.0 - 2022-09-09

Added

  • Added methods to save and read PLIERResults objects to disk using h5py, which should be dramatically faster than using json
  • Added a __getitem__ method to the PLIERResults class

Changed

  • Renamed PLIERResults.to_disk() to PLIERResults.to_json() and PLIERResults.from_dist() to PLIERResults.read_json()

1.4.0 - 2022-09-02

Fixed

  • Better handling for NaNs and Inf values in the normalized data when running plier()
  • Explictly list submodules as part of __all__ in __init__.py to hopefully solve certain submodules not being available when importing the main module

Misc

  • Update dependencies
  • Cleanup testing data

1.3.0 - 2022-08-22

Changed

  • revert newer (3.9+) style typing soas to reenable Python 3.8

1.2.0 - 2022-08-22

Changed

  • Simplified solveU

1.1.1 - 2022-08-17

Fixed

  • Should probably actually try using pyplier.plotting.plotTopZ before pushing.

1.1.0 - 2022-08-17

Fixed

  • pyplier.plotting.plotTopZ actually works now.

1.0.1 - 2022-08-15

Fixed

  • Replaced rich.console.Console with just using rich.print so as to avoid circular imports

1.0.0 - 2022-08-15

Changed

  • Reorganized the submodules to make a little more sense and so that not everything was plier.FUNCTION.FUNCTION; some functions changed to be PLIERRes methods

Removed

  • Removed multiple functions

0.14.0 - 2022-08-15

Fixed

  • In nameB, handle the situation where no pathways have a FDR less than the cutoff

0.13.1 - 2022-08-07

Removed

  • Removed a few unused or unnecessary dependencies to reduce the exposure to vulnerabilities

0.13.0 - 2022-07-31

Added

  • Unit test for pyplier.combinePaths.combinePaths()
    • also, while the results do not match those from R {PLIER}, pyplier's version is technically correct - PLIER's version has a bug

Fixed

  • Replaced inappropriate use of numpy.exp() where I should have used numpy.power() in pyplier.AUC.mannwhitney_conf_int()
  • remove stupid from typing import dict, list leftovers
  • pyplier.combinePaths.combinePaths() now returns an array of ints

0.12.0 - 2022-07-17

  • All apparently used functions implemented

Added

  • plotU() & VarianceExplained()

Changed

  • Changed minimum required Python version to 3.9

0.11.0 - 2022-07-04

Added

  • plotTopZallPath()

Fixed

  • plotTopZ() and solveU() now properly use PLIERRes members

0.10.0 - 2022-06-26

Added

  • plierResToMarkers, plotMat, and plotTopZ

0.9.0 - 2022-06-24

Added

  • Unit tests for the PLIERResults class
  • Start add use of structlog
  • added __eq__() method to the PLIERResults class

Changed

  • dev container now uses python 3.9
  • Replaced a few math functions in pyplier.AUC.mannwhitneyu_conf_int() with their numpy equivalents
  • pyplier.crossVal.crossVal() now sets the "pathway" column as the index for the summary dataframe
  • changed pyplier.PLIERResults.from_dict() method to a classmethod
  • gzipped several of the testing files

Fixed

  • Fixed column names for testing dataframes in test_solveU

0.8.0 - 2022-04-19

Added

  • New version of mannwhitney_conf_u(), which is a direct translation of the R code used in the portion of stats::wilcox.test() that generates the confidence interval for PLIER::AUC()
  • __repr__() method to the PLIERRes class

Changed

  • Simpler version of solveU() (at least temporarily)
  • Started using typechecked
  • Overhaul crossVal()

Fixes

  • plier() now runs and appears to do so essentially correctly; however, at current, it reports fewer identified LVs than {PLIER} does (about half as many)

Removed

  • Got rid of commonRows() as it was unnecessary

0.7.0 - 2022-02-08

Added

  • First draft of the main PLIER function, though it

Changed

  • Compress test data and change tests to reflect
  • Minor code changes to eliminate pytest warnings
  • Switched PLIERResults from a TypedDict to a class

0.6.2 - 2022-02-07

Changed

  • Removed numba as a dependency for now (it wasn't being used anyway), allowing the required numpy version to be updated to >1.21 (lower version had a vulnerability)
  • Update all dependencies

0.6.1 - 2022-02-06

Changed

  • Update dependenciess to eliminate ipython vulnerability

0.6.0 - 2022-02-06

Added

  • solveU() with unit test (though, no hypothesis-based unit tests yet)
  • A Docker devcontainer
  • Import functions into init to simplify intramodule use
  • Add icontract-based function input validation to some functions