v2.0.6

Minor updates:

• moved Metric.py from the environment package to the ml_metrics package
• updated YAML load(..) calls without specified Loader to safe_load(..) instead
• bugfix: removed caching from scikit-learn models
• updated cache location for multiple processes

v2.0.5

Minor bug fixes and usability improvements:

• added checks for encoder-ML compatibility when parsing
• added Moristia-Horn distance metric and improved HTML output for DistanceEncoder
• added LabelHelper for parsing labels
• added label checks before starting ComparisonData matrix computation
• removed sequence_weight requirement from FullSequenceImplanting YAML (it was not used)
• improved docs

v2.0.4

Minor bugfix: updated SequenceMetadata object to convert 'nan' for RegionType to None (manifested before as an error on loading existing ImmuneML-formatted dataset from disk).

v2.0.3

• standardized import/export metadata for repertoire datasets to always have field "identifier" instead of mixing it with "repertoire_identifier"

v2.0.2

• integrate with Zenodo to make the code citable

v2.0.1

Updates in v2.0.1 include minor bug fixes:

• fix path handling and serialization for ImmuneML format
• fix exported metadata for datasets with implanted signals
• minor updates related to Galaxy API

v2.0.0

Updates in version 2:

• datasets are now stored using binary numpy and YAML files and do not rely on pickle anymore;
• Pickle import and export is replaced with ImmuneML import and export where datasets are represented as YAML files, while the content (repertoire data, receptors, sequences) are stored in numpy files;
• 'store_encoded_data' option is removed from instructions; to store the encoded data, use DesignMatrixExporter report.

The package is also available at: https://pypi.org/project/immuneML/2.0.0/