NumericalFeature

class NumericalFeature[source]

Bases: object

Utility feature engineering class to process and filter numerical data structures, such as amino acid scales or a feature matrix.

It provides numeric helpers for the CPP feature engineering pipeline: extending the amino acid alphabet of a scale DataFrame, slicing per-residue tensors into sequence parts (numerical analog of SequenceFeature.get_df_parts()), reconstructing the model matrix X from CPP.run_num()-selected features (feature_matrix()), and removing redundant features by Pearson correlation.

Added in version 0.1.3.

Methods

`extend_alphabet`(df_scales, new_letter[, ...])	Extend amino acid alphabet of `df_scales` by new letter.
`feature_matrix`(features, dict_num_parts, ...)	Create the numerical-mode feature matrix `X` for given feature ids and per-part tensors.
`filter_correlation`(X[, max_cor])	Filter features based on Pearson correlation.
`get_parts`(df_seq, dict_num[, list_parts, ...])	Prepare Comparative Physicochemical Profiling (CPP) numerical-mode inputs by slicing sequences AND per-residue tensors with shared boundaries.