Geometrical Eigen-subspace Framework Based Molecular Conformation Representation for Efficient Structure Recognition and Comparison

摘要

We have developed an extended distance matrix approach to study the moleculargeometric configuration through spectral decomposition. It is shown that thepositions of all atoms in the eigen-space can be specified precisely by theireigen-coordinates, while the refined atomic eigen-subspace projection arrayadopted in our approach is demonstrated to be a competent invariant instructure comparison. Furthermore, a visual eigen-subspace projection function(EPF) is derived to characterize the surrounding configuration of an atomnaturally. A complete set of atomic EPFs constitute an intrinsic representationof molecular conformation, based on which the interatomic EPF distance andintermolecular EPF distance in the eigen-space can be reasonably defined.Exemplified with a few cases, the intermolecular EPF distance shows exceptionalrationality and efficiency in structure recognition and comparison.