Hash codes for the identification and classification of molecular structure elements


  • Wolf Dietrich Ihlenfeldt,

    1. Department of Knowledge-Based Information Engineering, Toyohashi University of Technology, Tempaku, Toyohashi 441, Japan
    Search for more papers by this author
  • Johann Gasteiger

    Corresponding author
    1. Institut für Organische Chemie, Technical University Munich, 85748 Garching, Germany
    • Computer-Chemie-Centrum, Universität Erlangen-Nürnberg, Nägelsbachstr. 25, D-91052 Erlangen, Germany
    Search for more papers by this author


A set of algorithms is presented which establish the topological identity of atoms, bonds, molecules, and ensembles of molecules from a basic connection table. The computationally inexpensive result is a fixed-length hash code which is suited for database applications and structure manipulation programs. The degree of differentiation between structural entities is adjusted easily for stereocenters, isotope labeling, atomic charges, and ionization locations or other properties. Special algorithms are presented which deal with problematic cases of uniform atomic environments. A number of practical applications demonstrate the usefulness of these hash codes. © 1994 by John Wiley & Sons, Inc.