0.1.3-SNAPSHOT

Merge branch 'main' of github.com:reality/klarigi

0.1.2

• More information in
• --classify function repaired
• --classify-with-variables with --classify now uses nExclusion of training set passed with --data, since obviously we shouldn't know the distribution of class balance in the test set. Ironically this tends to actually improve the performance, because because the training set is probably larger, the nExclusion scores are more refined w.r.t label...

Full Changelog: 0.1.1...0.1.2

0.1.1-2

• Fix for --max-inclusion and friends on CalculateOI

Full Changelog: 0.1.1...0.1.1-2

0.1.1

• Major fix: Previously, overall inclusion calculation was using only exact matching, which is not the correct implementation of the scoring system. This has been fixed.
• Performance and efficiency improvements, particularly surrounding classification. Use of pre-scoring values speeds things up a lot. Some multiprocessing stuff in data loading. Repairs for potential race conditions.
• Other small output/usability fixes

Full Changelog: 0.1.0...0.1.1

0.1.0

This marks the release of 0.1.0, (essentially) the version used for the paper.

• Many output fixes, correct LaTeX things
• Replace ECM with UCM (so ECM is the default classification mode)
• --classify-with-variables
• Redefinition of inclusion/exclusion scores
• Power -> r-score
• Many bug-fixes
• Much much more
• Introduce pre-scoring for greater efficiency
• Various speed/efficiency improvements
• Usability/avoidance of stack trace captures
• --show-warnings required for IC warnings

Full Changelog: 0.0.12...0.1.0

0.1.0-RC-2

Full Changelog: 0.0.13-SNAPSHOT-2...0.1.0-BETA

0.0.13-SNAPSHOT-2

pval fix, and --include-all mode for scoring

0.0.13-SNAPSHOT

• permutation testing to provide p-values for inclusion and
• redefinition of exclusion score to be proportion of total explained entities are internal to group
• exclusive-group-mode
• exclude-group-litter
• phenopacket support initial
• output improved for cases with single groups (don't print exclusion etc)
• --min-power

0.0.12

• Enable provision of groupings for similarity calculations
• Fixes for the algorithm that previously caused no results when total coverage is low
• --debug output for StepDown algorithm

0.0.11

• multiple group membership delimited by ; in the group section of the inputs
• --threads [int] option for multiprocessing
• rewriting datastructures for time efficiency
• bug fixes
• data frame output
• --power mode