Beta Features

Some AAanalysis tools are still under active development and are marked beta. For a beta tool, the API (signatures, defaults, return objects) may change between minor releases without the usual deprecation cycle, so pin a version (e.g. pip install aaanalysis==1.1.0) if you depend on the current behaviour. Everything else in the public API keeps the ordinary deprecation cycle.

Each tool below repeats this warning in its own documentation. The list is generated from the code: a symbol appears here exactly when its docstring carries the **Experimental.** warning, and a drift test keeps this page in sync, so it cannot go stale.

An entry written ap.<name> is a golden pipeline reached through import aaanalysis.pipe as ap; every other entry is a class reached through import aaanalysis as aa.

Tool	Purpose	Added in
SequenceFeatureTransformer	Leak-free CPP feature selection as a scikit-learn transformer.	1.1.0
SeqOpt	Multi-objective directed evolution over sequence variants.	1.1.0
SeqOptPlot	Pareto-front and convergence plots for SeqOpt results.	1.1.0
AAPred	Evaluation and deployment of sequence-based prediction models.	1.1.0
AAPredPlot	Evaluation and prediction figures for AAPred results.	1.1.0
ReliabilityModel	Per-prediction trust: calibration, uncertainty, and applicability domain.	1.1.0
ReliabilityModelPlot	Calibration and trust-axis plots for ReliabilityModel outputs.	1.1.0
ModelEvaluator	Cross-validated evaluation and paired comparison of models.	1.1.0
CPPStructurePlot	CPP feature impact painted onto a 3D protein structure.	1.1.0
ap.find_features	Staged CPP AutoML search for a discriminating feature set.	1.1.0
ap.predict_samples	Training and comparison of predictors across feature sets and models in one call.	1.1.0
ap.explain_features	Per-sample SHAP impact and the SHAP-coloured feature map.	1.1.0