doubletdetection v4.2

Fixes a bug in 4.0, where default louvain used directed=True, when it should use False.

doubletdetection v4.1

Merge pull request #151 from JonathanShor/dev_4.1

Dev 4.1

doubletdetection v4.0

Merge pull request #139 from JonathanShor/cluster_opts

Dev 4.0

doubletdetection v3.0

• Distributed via PyPI
• Remove unnecessary plotting code

HOTFIX: Correct setup.py installation

Fix an misconfiguration in setup.py.
Improved README while we're at it.

HOTFIX: Correct Defaults & Docstrings

Minor corrections to align parameter defaults with documentation around use_phenograph and standard_scaling parameters.

Sparse Scanpy Speedups

• Improves installation process through one pip install command and removes requirements.txt
• Adds documentation
• Uses sparse matrices for raw counts in the backend for speedups
• Uses scanpy for log normalization and PCA and optionally clustering
• New use_phenograph option to disable PhenoGraph and use Louvain as implemented in scanpy (much faster)
• Adds binary verbose option. Set to True to turn on informational messages (previous default behavior).
• Replaces tsne plot with umap and no longer clusters data, just visualizes doublets on the umap
• With new scanpy integration, doubletdetection requires at least Python 3.6.

DOI

Minor release updating the front README to reflect the work's DOI.

Beauty from Random States

• Adds random state parameter for PCA, tSNE, synthetic doublet creation in BoostClassifer - closes #106
• Adds random state parameter for PCA, tSNE in tsne plot
• tsne plot returns PhenoGraph communities - closes #109
• tsne plot accepts array-like through sklearn check array - closes #104
• improves readability of threshold plot - closes #102
• notebook uses voter threshold of 0.8 (non-default)

Log p-values.

We noticed floating point rounding errors were masking distinction between very high p-values that were in fact different. The method now calculates and works with the log p-values internally, and makes them available via the all_log_p_values_ attribute.

Due to the problems with the un-logged p-values, all_p_values_ is deprecated, and will be removed in v3.0. We encourage all users to avoid using it, and instead exponentiate all_log_p_values_ (with care!) as needed.

To help understand the significance of different p and voter thresholds, we have added plot.threshold to provide a visualization of their interaction.