Fast taxonomy of chemical structures selected from IR spectral database

The results of taxonomy of organic structural formulas selected by a retrieval system from an infrared (IR) spectral database are presented. Two models of molecular graph are compared: connectivity matrix and vector description (exhaustive set of nonisomorphic k-vertex fragments). It is shown that the taxonomy of structures based on the shortest spanning tree of distance graph unites structurally similar compounds in groups (taxons). The taxonomy of structural formulas was performed for more than 100 spectral search results. The vector description of graphs provides fast taxonomy of structural formulas of selected compounds and makes easier the recognition of large structural fragments of the query compound.