Polyclonal

Latest version: v6.12

Safety actively analyzes 681857 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

Page 4 of 7

4.0

---
- Enable (and activate by default) fitting of a Hill coefficient and non-neutralized fraction in the curves. Previously, the Hill coefficient had been constrained to one and the non-neutralized fraction to zero. This is a **major change** that will alter the results of fitting models. To get the old behavior, call ``Polyclonal.fit`` with ``fix_hill_coefficient=True`` and ``fix_non_neutralized_frac=True``. Specific changes associated with this update include:
- Add ``hill_coefficient_df`` and ``non_neutralized_frac_df`` as parameters and properties of ``Polyclonal`` and ``PolyclonalCollection``.
- Restructure internal organization of ``Polyclonal._params`` (this was private, so external code should not be using anyway).
- Add regularization to Hill coefficient and non-neutralized fraction to ``Polyclonal.fit``.
- Add the "two-epitope" example to illustrate the non-one Hill coefficient and non-neutralized fractions not equal to zero.
- Update ``reg_escape_weight`` to 0.05 in ``Polyclonal.fit`` because it seems like many people in lab were using larger values. This will **change results** of fitting models because old default was 0.02.
- Add the ``curves_plot`` and ``curve_specs_df`` to ``Polyclonal`` / ``PolyclonalCollection``, ``curves_plot`` to the ``plots`` module, and prefer use of this over the activity barplots in the docs. The reason is that this shows the activity, Hill coefficient, and non-neutralized frac.
- Remove bootstrapping from docs as this isn't really a recommended procedure.
- Updated the default colors for epitopes (changing fourth to light blue and adding olive as fifth)

3.4

---
- Plotting improvements:
- Add ``heatmap_max_fixed`` and ``heatmap_min_fixed`` to ``plot.lineplot_and_heatmap``
- Make it easier to access per-model measurements for averages of models:
- added ``PolyclonalCollection.unique_descriptor_names`` attribute.
- added ``PolyclonalCollection.mut_escape_df_model_values`` property
- added ``per_model_tooltip`` option to ``PolyclonalCollection.mut_escape_plot``, and make this the default when <=5 models.
- Heatmap tooltips now show missing (null) numerical values as *NaN* rather than as 0.
- Update ``altair`` version to commit to *f8912bad75d4247ab726b639968b13315161660a* (note that the latest version still not merged on ``altair`` main branch, so still having to install from GitHub). In this new version, ``alt.add_parameter`` becomes ``alt.add_params`` and ``alt.parameter`` becomes ``alt.param``.
- Implemented slider that hides rather than filters mutations on the heatmaps. This is designed for mutation effect filtering where we'd like to be able to see which mutations have poor effects. They are now filtered from lineplot calculation and shown as gray. Adds the following parameters to ``plot.lineplot_and_heatmap``:
- ``addtl_slider_stats_hide_not_filter``
- ``hide_color``
- also very slight changes to colors in background of heatmap.
- Re-order default positive color scheme to put green before dark byzantium to give better clarity relative to hidden / filtered values in heatmaps.
- Updated notebooks to use new plotting.

3.3

---
- Add options for lineplot only or heatmap only in ``plot.lineplot_and_heatmap`` (these are ``show_zoombar``, ``show_lineplot``, and ``show_heatmap``).
- Add ``scale_stat_col`` option to ``plot.lineplot_and_heatmap``.
- Add ``rename_stat_col`` option to ``plot.lineplot_and_heatmap``.

3.2

---
- Allow non-integer PDB numbers in B-factor re-assignment in ``reassign_b_factor``.

3.1

----
- Change default positive colors.

3.0

----
- Make ``Polyclonal.spatial_distances`` a public attribute.
- ``Polyclonal.fit`` allows epitopes with identical activities if they have different escape.
- Activity regularization penalizes both positive and negative values.
- Checks and int versus str dtype adjustment for ``spatial_distances`` in ``Polyclonal``.
- Adjust activity regularization based on concentration geometric mean so it is not sensitive to units of concentration.
- Renamed what was previously called the epitope similarity regularization to epitope uniqueness-squared, as it's a second uniqueness regularization but operates on square of escape values.
- Change default regularization strengths in ``Polyclonal.fit``.
- Rename ``reg_spatial_weight2`` to ``reg_spatial2_weight`` for ``Polyclonal.fit``.
- Update default values of regularization weights.
- Added antibody cocktail and HIV serum examples.
- Updated examples, for instance by adding spatial regularization to RBD example and slightly changing simulated data.

Page 4 of 7

© 2024 Safety CLI Cybersecurity Inc. All Rights Reserved.