HSQC spectral based similarity matching of compounds using nearest neighbours and a fast discrete genetic algorithm

HSQC spectra are routinely acquired for chemical structure analysis based on hydrogen and carbon chemical environments. Two fast HSQC peak matching algorithms have been developed; a nearest neighbour approach and a probabilistic method based on an existing discrete genetic algorithm. Both of these techniques are intended to find HSQC spectra matches that supplement information generated by established molecular fingerprint methods. Our results are compared to those calculated using a specific implementation of molecular fingerprints. The nearest neighbour and genetic algorithm-based methods ranked highly particular structures missed by molecular fingerprints. Our analysis shows that by complementing molecular fingerprint matches with our findings, a comprehensive list of matches can be identified. The refined list of compounds could be used to improve the quality of compounds used in screening libraries in the pharmaceutical industry.