Chemical substructure identification by mass spectral library searching.

Area of Science:

• Analytical Chemistry
• Computational Chemistry

Background:

• Accurate identification of unknown compounds is crucial in chemistry.
• Electron-ionization mass spectrometry (EI-MS) is a powerful tool for compound identification.
• Existing library-search methods have limitations in accuracy and efficiency.

Purpose of the Study:

• To develop an optimized library-search procedure for identifying structural features of unknown compounds from EI-MS data.
• To improve upon existing K-nearest neighbor methods by incorporating enhanced screening and weighting schemes.

Main Methods:

• A novel library-search procedure was developed, weighting retrieved spectra by similarity.
• Included a "peaks-in-common" screening step to reduce search times.
• Utilized an optimized dot product function for match factor calculation and incorporated "neutral loss" peaks for improved substructure identification when molecular weight is known.

Main Results:

• Correlations between substructure presence and library retrievals were established using the NIST/EPA/NIH library and a 7891 compound test set.
• The method allows for the estimation of probabilities for substructure occurrence and absence in unknown compounds.
• The developed method demonstrated improvements over traditional K-nearest neighbor approaches.

Conclusions:

• The optimized library-search procedure offers enhanced accuracy and efficiency for EI-MS-based compound identification.
• The method provides probabilistic estimations of substructure presence, aiding in structural elucidation.
• This approach represents a significant advancement in computational spectral analysis for unknown compound identification.