Similarity search for a 1H-NMR spectroscopic data base

摘要: A similarity-search system is described for proton-NMR spectroscopy. In order to achieve fast retrieval of reference compounds, the 1H-NMR spectra data base and unknown are encoded in a bitsring. The individual bits binary signature describe different features spectra. Part coupling information coded such way that effects magnetic field strength taken into account. encoding thus permits search identical structurally similar compounds even when were recorded at strengths. Because consists weighted comparison bits, each them describing spectral features, choice kinds searches possible with same by selecting appropriate weight vectors. Thus specific spectroscopic can be selected search. Such context-sensitive allows, example, having multiplicities or subspectra given (e.g., aromatic) region spectrum. Furthermore, adjusting two “software knobs” which influence normalization results, user choose between extremes forward reverse search, an identity similarity classification results tested on small library containing 550 including some mixtures duplicates under experimental conditions 250 400 MHz.