AAontology: Classification of Amino Acid Scales

Amino acid scales turn each residue into a number (hydrophobicity, charge, size, and so on), but hundreds of scales exist and most of them measure overlapping properties. AAontology is a two-level classification that organises these scales by what they actually capture, introduced in [Breimann24b]. It is what lets a CPP feature carry a readable physicochemical meaning rather than an opaque scale identifier: every scale belongs to a subcategory (for example Polarity) and a top-level category (for example ASA/Volume), so a feature signature can be summarised by category and interpreted at a glance.

Provided by

In AAanalysis the ontology ships with the data. load_scales() returns the scales (df_scales) together with their two-level classification (df_cat). Browse the full category and subcategory list in Data Tables, and see the API reference for the loader options.

AAontology was created by automatic scale classification followed by manual refinement:

../../_images/scheme_AAontology1.png — Workflow of scale classification from [Breimann24b].

It comprises 586 amino acid scales, organised into 67 subcategories and 8 categories. Grouping scales this way keeps the feature space interpretable: instead of ranking 586 raw scales, CPP can pick representatives per category (see AAclust) and report a signature in terms of a handful of physicochemical themes.