Collision Cross Section Prediction Based on Machine Learning.

Area of Science:

• Analytical Chemistry
• Computational Chemistry
• Biochemistry

Background:

• Ion mobility-mass spectrometry (IM-MS) offers advanced separation for complex samples.
• Lack of reference standards hinders comprehensive chemical characterization.
• Machine learning (ML) integration with IM-MS presents a solution for data analysis and database creation.

Purpose of the Study:

• To review advances in collision cross-section (CCS) prediction using ML over the past two decades.
• To compare different ion mobility technologies and their principles.
• To highlight ML-based CCS prediction procedures and theoretical calculations.

Main Methods:

• Summary of ML algorithms and techniques for CCS prediction.
• Description of various ion mobility-mass spectrometry instrumentation.
• Explanation of data acquisition, variable optimization, and model evaluation in ML-based CCS prediction.
• Inclusion of quantum chemistry, molecular dynamics, and theoretical CCS calculations.

Main Results:

• ML-driven CCS prediction facilitates the creation of extensive CCS databases.
• This approach enables rapid, comprehensive, and accurate characterization of chemical components.
• Overcomes limitations posed by the absence of chemical reference standards.

Conclusions:

• ML-enhanced IM-MS is a powerful tool for analyzing complex biological and natural product samples.
• The development of CCS prediction models is crucial for advancing metabolomics, natural product research, and food analysis.
• This review provides a comprehensive overview of ML applications in IM-MS for chemical characterization.