v2.0.1 (Minor fixes)

Minor bug fixes addressing issues: #21, #22, #23, #24

v2.0.0

CytoPy version 2.0.0 is the mature release of CytoPy with a new design that focuses on persistent data structures that can be shared between multiple methodologies for identifying cell populations in cytometry data. CytoPy 2.0 is not compatible with previous releases.

New features include:

• Updated ODM design
• Support for both FCS files and cytometry data in other formats that can be read into a Pandas DataFrame
• Updated autonomous gating to include hyperparameter search and local normalisation
• Integration of Harmony for correcting batch effect
• Decoupling of supervised learning from the database and expansion to support any supervised classifier from the Scikit-Learn ecosystem
• Expansion of feature extraction and selection techniques

Clustering fix

This release corrects some major errors encountered in the flow.clustering module that was preventing clusters from being saved to the database and retrieved correctely.

In addition test coverage has been increase and some minor bugs fixed.

Addition of the flow.gate_search module for hyperparameter tuning of gating parameters. Still experimental and docs not available yet.

First major release v1.0.0

Version 1.0.0

This is the first major release of CytoPy following the early release of v0.0.1 and updated in v0.0.5 and v0.1.0. This first major release includes fundamental changes to data management and therefore is not backward compatible with previous versions.

Improvements in the new version are:

• Dis-coupling of single cell data with the MongoDB database, allowing for easy migration of large data stores to external drives, cloud drives, etc
• Redesign of the autonomous gating system with a completely algorithm agnostic focus; the user can access any Scikit-learn algorithm for the purposes of gating
• Improvements to the similarity matrix and the use of a fast convolution based kernel density estimation function (KDEpy) to reduce compute time
• Expansion of supervised classification to provide access to any Scikit-Learn classifier or Keras model
• Expansion of high-dimensional clustering to provide access to any Scikit-Learn model from the cluster or mixture modules. Improved access to FlowSOM algorithm
• Simplification of data storage for clusters improving the ability to retrieve statistics in the feature_extraction module