UCLA-DOE Laboratory of Structural Biology and Molecular Medicine, Molecular Biology Institute, UCLA, Los Angeles, CA 90095-1570
3D-1D profile methods aim to assign protein sequences to folding classes, or to detect sequence similarities, by observing the probability of occurrence of certain protein properties and their variation by residue type. Previously studied properties include buried side chain area and fraction of the side chain in contact with polar atoms (Bowie, Luethy and Eisenberg, Science v. 253, p. 164, 1991).
Directional profiles use properties tabulated in several directions in some residue-based coordinate system, for example, a count of non-polar atoms in four lobes centered on the alpha carbon and pointing along the directions of its four bonds. Generalizations involve substituting other, unrelated, properties for those in statistically less important directions.