Method for assessing the statistical significance of mass spectral similarities using basic local alignment search...

Area of Science:

• Analytical Chemistry
• Bioinformatics
• Metabolomics

Background:

• Metabolomics relies on mass spectral similarity for metabolite identification.
• Raw metabolome data contains noise, leading to false positives in metabolite identification.
• Accurate statistical assessment of mass spectral similarity is crucial for reliable metabolomics.

Purpose of the Study:

• To develop a novel method for statistically assessing mass spectral similarity.
• To improve the accuracy of metabolite identification in gas chromatography/mass spectrometry-based metabolomics.
• To reduce the false discovery rate in metabolomics data analysis.

Main Methods:

• Developed an electron ionization (EI) mass spectrometry-BLAST method using modified Karlin-Altschul statistics.
• Created a scoring scheme to calculate similarity scores and P values between mass spectra.
• Utilized Monte Carlo simulations to validate the statistical model and expected number of hits (E value).

Main Results:

• The developed method accurately assesses statistical significance of mass spectral similarities.
• Analysis of green tea extract metabolome data identified 93 out of 171 signals with P < 0.015.
• Estimated false discovery rate was 2.8%, indicating a reasonable search threshold for metabolite identification.

Conclusions:

• The novel EI mass spectrometry-BLAST method provides a statistically robust approach for metabolite identification.
• This method significantly enhances the reliability of metabolomics data by minimizing false positives.
• The findings support the use of P < 0.015 as a reliable threshold for metabolite identification in metabolomics studies.