Collision Cross Section Prediction Based on Machine Learning

Abstract

Ion mobility-mass spectrometry (IM-MS) is a powerful separation technique providing an additional dimension of separation to support the enhanced separation and characterization of complex components from the tissue metabolome and medicinal herbs. The integration of machine learning (ML) with IM-MS can overcome the barrier to the lack of reference standards, promoting the creation of a large number of proprietary collision cross section (CCS) databases, which help to achieve the rapid, comprehensive, and accurate characterization of the contained chemical components. In this review, advances in CCS prediction using ML in the past 2 decades are summarized. The advantages of ion mobility-mass spectrometers and the commercially available ion mobility technologies with different principles (e.g., time dispersive, confinement and selective release, and space dispersive) are introduced and compared. The general procedures involved in CCS prediction based on ML (acquisition and optimization of the independent and dependent variables, model construction and evaluation, etc.) are highlighted. In addition, quantum chemistry, molecular dynamics, and CCS theoretical calculations are also described. Finally, the applications of CCS prediction in metabolomics, natural products, foods, and the other research fields are reflected.

Ion mobility calculations of flexible all-atom systems at arbitrary fields using two-temperature theory

“…It is also quite impossible at this time for machine learning algorithms to reproduce results of structure-varying and high-field calculations. 66–68 Interpreting the results, it is difficult to ascertain the impact of inelasticity on the mobility of polyatomic ions in polyatomic gases at this time. When considering the expansion of structures due to heating, it becomes evident that the contribution of energy directed into the gas's internal degrees of freedom is somewhat less than initially anticipated based on the raw data.…”

Collision cross section measurement and prediction methods in omics

“…The external dataset is an independent set of compounds or CCS values that are not included in the training and internal datasets 51 . Recently, Li et al reviewed the type of available MDs, available MDs' computing software, preprocessing and optimization steps of MDs, and different ML prediction algorithms and platforms in detail 145 . In the next sections, a summary of the different types of ML methods and their applications is given (Table 3).…”

Predicting the Predicted: A Comparison of Machine Learning-Based Collision Cross-Section Prediction Models for Small Molecules

“…Some models like DeepCCS 9 directly encode the Simplified Molecular Input Line Entry System (SMILES) as binary matrices for molecular representation. A combination or a selection of these features are then used as an input to train different ML paradigms, which have included support vector regression, 10 random forest, 7 and artificial neural networks. 9,11 Some of these models have been published alongside small or relatively large CCS libraries, usually covering hundreds or a few thousands of molecules, with AllCCS being one of the largest libraries available so far.…”

“…In-silico prediction of CCS values commonly relies on chemical first-principles 6 or data-driven approaches such as ML. 7 ML approaches tend to be faster and since ML-based predictions are heavily dependent on the quality and the structural diversity of the input data, there is a concerted effort to build CCS databases that can eventually be used for the robust prediction of CCS values. 8 Accordingly, different ML frameworks and tools have been implemented for the prediction of CCS values, which are trained on curated experimental data sets.…”