Comparison of algorithms and databases for matching unknown mass spectra

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Computational Chemistry

Background:

• Electron-ionization mass spectrometry is crucial for chemical identification.
• INCOS and Probability Based Matching (PBM) are leading algorithms for spectral analysis.
• Evaluating algorithm performance on large spectral databases is essential.

Purpose of the Study:

• To compare the performance of INCOS and PBM algorithms for identifying unknown electron-ionization mass spectra.
• To assess algorithm accuracy with varying sample purity and database sizes.
• To evaluate the effectiveness of these algorithms in identifying mixtures.

Main Methods:

• Utilized the National Institute of Standards and Technology (NIST) and Wiley spectral databases.
• Tested INCOS and PBM algorithms against a database of 62,235 and 228,998 spectra, respectively.
• Assessed algorithm accuracy using different matching criteria and spectrum purities (high purity and 85% purity).

Main Results:

• For high purity unknowns against the NIST database, both algorithms achieved approximately 75% correct rank 1 answers.
• Against the larger Wiley database (50% retrieval), PBM (54% correct) outperformed INCOS (42% correct).
• For 85% purity unknowns against Wiley, PBM accuracy was 48% versus INCOS at 27%.
• In a two-compound mixture, PBM identified both components within the top ten answers, while INCOS did not identify either within the top three.

Conclusions:

• PBM demonstrates superior performance over INCOS, particularly for complex mixtures and larger spectral databases.
• Algorithm choice significantly impacts the accuracy of mass spectral identification.
• PBM offers a more robust solution for identifying components in mixed spectra.