T. F. Bogdanova

Taxonomy of Structures Selected from the IR Spectroscopy Database

This paper reports on taxonomy of the structural formulas of organic compounds selected from an IR spectroscopy database (DB) based on the similarity between the query spectrum and DB spectra. Two molecular graph models are compared: connectivity matrix and vector representation (exhaustive set of nonisomorphic connected k-vertex fragments). In both cases, taxonomy according to the shortest distance framework of the distance graph gives a classification of structural analogs into groups (taxons). An analysis of IR spectra and common subgraphs in the respective groups of graphs reveals large structural fragments of the compounds.

Interpretation of IR spectra using complete sets of structure fragment compositions and an IR search system. 3. Analysis of large fragments

The results of identification of five-, six-, and seven-node fragments by the IR spectra of compounds are characterized by quantitative parameters: probability, reliability, and types of fragments. This analysis is carried out using the database containing more than 11,000 complete IR spectra and structures of organic compounds. It is analyzed how the fractions of correct fragments in the search result depend on the threshold occurrence and nonrandomness of fragment selection. Examples of statistical reliability of the revealed fragments are given.

Interpretation of IR spectra using complete sets of fragment compositions and an ir search system. 2. Determination of microfragment compositions of organic compounds

Identification of fragments of organic compounds by the IR spectra of the latter using complete sets of fragments and an IR database of more than 11,000 complete spectra and structures of organic compounds is treated. Probability and reliability of identification are estimated, and types of revealed fragments are examined. For many fragments, automatic identification is possible. The influence of the threshold and nonrandom occurrences of the fragment in the resulting list of structures on the efficiency of identification are considered. The reliability of identification is justified statistically.

Applicability of a mass spectrum-fragment composition database to structure elucidation of compounds

Representation of structures as complete sets of nonisomorphic k-vertex subgraphs (fragments) in mass spectrometric databases offers some new possibilities in structure elucidation of compounds by their mass spectra (determination of fragment composition). The results of searches by mass spectra of 300 compounds with known structures are used as an example to analyze the dependence of correct recognition of fragments on their threshold nonrandom occurrence in the resulting list of structures.

Complete sets of fragment compositions of structures for ir spectrum interpretation using a database. 4. Forming the most probable hypothesis about the structure of the unknown

This paper offers a new approach to forming hypotheses about the structure of organic compounds. The approach uses the fragment compositions of the structures selected by seeking analogs of the IR spectrum of the compound in the database. Even with “noise” fragments, the whole set of revealed fragments may be used for generating the possible structures of the unknown due to the variety of intersecting fragments in the selected structures. Forming the most plausible hypothesis about the structure of the compound is treated in detail, and the results are shown on particular examples.

COMPLETE SETS OF STRUCTURE FRAGMENTS FOR INTERPRETING IR SPECTRA USING IR DATABASES

This paper studies the possibility of revealing various fragments (not given beforehand) in a structure by analyzing the structures selected from the database as a result of a retrieval by the IR spectrum of the compound; the fragments range from those with two nodes (e.g., C=O, C=C) to those with some limited number of connected nodes. It is shown that complete fragment sets reflecting the composition of the selected compound (fragment compositions) contain the information about the set of bonded fragments of the compound. Various factors affecting analysis of the compiled list of fragments are considered using a particular example. Different techniques for revealing structural information are discussed.