Several descriptors applied to peptide structure-activity and/or structure-
property relationships have been developed in recent years. This report des
cribes new descriptors for the natural amino acids which have been derived
from the principal component analysis (PCA) applied on the MS-WHIM 3D-descr
iption matrices. MS-WHIM indexes are a collection of 36 statistical indexes
aimed at extracting and condensing steric and electrostatic 3D-properties
of a molecule. These new descriptors have been developed both on extended s
ide-chain conformation and on rotamer library of natural amino acids. The m
ethod appeared to be more potent when describing a single conformation (i.e
. extended) than when applied collectively on library conformation families
. MS-WHIM scores, however, were shown to efficacely describe and correctly
classify the natural amino acid features and to provide sound statistical m
odels either predicting activities of two peptide sets taken as a test or c
orrelating amino acid chemicophysical properties, like water-accessible sur
face area, to general structural features of amino acids.