NEURAL-NETWORK MODEL OF THE GENETIC-CODE IS STRONGLY CORRELATED TO THE GES SCALE OF AMINO-ACID TRANSFER FREE-ENERGIES

A neural network trained to classify the 61 nucleotide triplets of the genetic code into 20 amino acid categories develops in its internal r epresentation a pattern matching the relative cost of transferring ami no acids with satisfied backbone hydrogen bonds from water to an envir onment of dielectric constant of roughly 2.0. Such environments are ty pically found in lipid membranes or in the interior of proteins. In le arning the mapping between the codons and the categories, the network groups the amino acids according to the scale of transfer free energie s developed by Engelman, Goldman and Steitz. Several other scales base d on internal preference statistics also agree reasonably well with th e network grouping. The network is able to relate the structure of the genetic code to quantifications of amino acid hydrophobicity-hydrophi licity more systematically than the numerous attempts made earlier. Du e to its inherent non-linearity the code is also shown to impose decis ive constraints on algorithmic analysis of the protein coding potentia l of DNA.