Releases: GenoML/genoml2
v1.5.4 - New pip version
Several key updates, including:
- Addition of preliminary multiclass prediction capabilities (which is still under active development)
- Refinement of continuous module
- Bug fixes, both big and small
- Changes to munging and harmonization workflows to allow train and test data to be munged together upfront and ensure they are preprocessed under the same conditions
- File restructuring for increased modularity
- Addition of log file in outputs for reproducibility
thank you @spencermg for implementing these improvements and @nvk23 for feedback!
v1.0.1 - Updates
Updates to packages, README, and minor bug fixes
v1.0.0-beta.11
This release includes
- Fix to issue #25
- Other minor bug fixes
- Updates to the requirements file
v1.0.0-beta.10
This release includes the "training wheels" to prevent nominating overfitting and also includes new probability plots across all supervised ML scripts
v1.0.0-beta.9
Changes in discrete training to nominating the best algorithm and the probability plots
v1.0.0-beta.8
Minor, but necessary, changes
v1.0.0-beta.7
v1.0.0-beta.6
This release has addressed a few issues:
- Issue #12: [Experimental] fix to using tuned models to test unseen, harmonized data
- Issue #16: Users now have more control to adjusting without reducing if desired, and choosing to normalize after adjusting if desired
- Issue #17: Users can now control stringency of r^2 value when pruning with PLINK using the
--r2_cutoff
The README has been updated to reflect these changes better, please report any issues to the GitHub issues page!
v1.0.0-beta.5
Implementing a "global adjuster" - flags in the munging process to account for additional covariates and confounders you would like to adjust for. This comes from feature request/issue #11 - Adjust genotypes for tSNE/UMAP/PC loadings.
v1.0.0-beta.4
Addressing issue #13 - next release might not be for a while!