AAclust: Selecting Redundancy-Reduced Scale Sets

AAclust (Amino Acid clustering) is a clustering wrapper framework for selecting redundancy-reduced sets of amino acid scales, introduced in [Breimann24a]. Using Pearson correlation, AAclust optimizes the number of clusters (k) and selects one representative scale per cluster, as illustrated in the figure below:

Scheme of AAclust algorithm with clustering of amino acid scales and scale selection, from [Breimann24a].

AAclust introduces two modes for defining the number of clusters (k):

• k-optimized: AAclust automatically optimizes k, streamlining the scale selection process.

• k-based: The user specifies k, allowing for custom configurations.

The distinctions between these modes and their respective AAclust options are depicted below:

Operational modes of AAclust for determining the number of clusters, from [Breimann24a].

Generally, AAclust works not only with amino acid scales sets but also with any set of numerical scales.

See our AAclust Tutorial for hands-on examples.