Protein Tetrahedral Networks by Invariant Shape Coordinates.

Representations of large molecules, typically proteins or DNA, for classification purposes, are commonly obtained by molecular networks. These are usually made up by a set of specific atoms (e.g. C α atoms in the amino acids) of the molecule plus a neighbourhood criterium, that establishes links between the centers, so building the network. The main objectve of such approaches is the discrimination of structures, the induction of grouping of them, depending on the basis of structural molecular properties. Here we propose applications of invariant shape coordinates as parameters for the construction of enhanced molecular networks for protein classification.