Bayesian peak tracking: A novel probabilistic approach to match GCxGC chromatograms

Area of Science:

• Analytical Chemistry
• Chromatography
• Chemometrics

Background:

• Gas chromatography-gas chromatography-flame ionization detection (GCxGC-FID) is a powerful separation technique.
• Accurate peak tracking is crucial for comparing complex sample analyses across different conditions.
• Existing deterministic peak tracking methods can be limited by their inability to handle uncertainty.

Purpose of the Study:

• To introduce a novel probabilistic peak tracking method using Bayesian statistics for GCxGC-FID data.
• To compare the performance of the proposed probabilistic method against traditional deterministic approaches.
• To develop algorithms for robust and efficient peak assignment in multi-dimensional chromatography.

Main Methods:

• Development of a Bayesian statistical framework for peak assignment between two GCxGC-FID datasets.
• Implementation of two algorithms: Blind Peak Tracking Algorithm (BPTA) and Peak Table Matching Algorithm (PTMA).
• Probabilistic quantification of peak matching uncertainty, ranking assignments by posterior probability.

Main Results:

• The Peak Table Matching Algorithm (PTMA) achieved 78% correct peak assignments for a diesel sample.
• The proposed probabilistic method offers a significant advantage over deterministic methods by quantifying matching uncertainty.
• PTMA demonstrated exceptional speed in processing GCxGC-FID chromatograms.

Conclusions:

• The novel Bayesian peak tracking method provides a probabilistic approach to peak assignment in GCxGC-FID.
• This method enhances the reliability of comparative analyses by managing uncertainty in peak matching.
• The developed algorithms, particularly PTMA, are efficient and accurate for complex sample analysis.