Predicting ion mobility collision cross sections and assessing prediction variation by combining conventional and...

Area of Science:

• Analytical Chemistry
• Computational Chemistry
• Molecular Modeling

Background:

• Collision cross section (CCS) values from ion mobility are crucial for molecular characterization in complex mixtures.
• Accurate CCS prediction aids in identifying and distinguishing between similar molecules, especially isobars.

Purpose of the Study:

• To develop a novel machine learning (ML) based method for predicting CCS values.
• To integrate molecular modeling (MM) with ML for enhanced CCS prediction accuracy.
• To improve the identification of compounds, particularly isobaric analytes, in complex mixtures.

Main Methods:

• A novel machine learning (ML) model was developed, integrating molecular modeling (MM) techniques.
• The ML model was trained and validated for predicting CCS values of singly charged small molecules.
• The correlation between charge localization, CCS prediction accuracy, and gas-phase proton affinity was investigated.

Main Results:

• The ML-based method accurately predicts CCS values for a wide range of small molecules.
• The model demonstrates superior performance compared to existing methods, improving the ranking and probability assignment for isobaric analyte identification.
• Charge localization was found to correlate with CCS prediction accuracy, suggesting a potential error proxy.

Conclusions:

• The developed ML approach offers a significant advancement in the accurate, computational prediction of CCS values.
• This method enhances the reliability of molecular identification in complex analytical samples.
• Findings pave the way for broader application of computationally derived CCS values in chemical analysis.