Graph-based machine learning interprets and predicts diagnostic isomer-selective ion-molecule reactions in tandem mass spectrometry.

Jonathan Fine,Hilkka I Kenttämaa,Xin Ma,Judy Kuan-Yu Liu,Gaurav Chopra,Victoria M Boulos,Kawthar Z Alzarieni,Armen Beck

doi:10.1039/d0sc02530e

Abstract

Diagnostic ion–molecule reactions employed in tandem mass spectrometry experiments can frequently be used to differentiate between isomeric compounds unlike the popular collision-activated dissociation methodology. Selected neutral reagents, such as 2-methoxypropene (MOP), are introduced into an ion trap mass spectrometer where they react with protonated analytes to yield product ions that are diagnostic for the functional groups present in the analytes. However, the understanding and interpretation of the mass spectra obtained can be challenging and time-consuming. Here, we introduce the first bootstrapped decision tree model trained on 36 known ion–molecule reactions with MOP. It uses the graph-based connectivity of analytes' functional groups as input to predict whether the protonated analyte will undergo a diagnostic reaction with MOP. A Cohen kappa statistic of 0.70 was achieved with a blind test set, suggesting substantial inter-model reliability on limited training data. Prospective diagnostic product predictions were experimentally tested for 13 previously unpublished analytes. We introduce chemical reactivity flowcharts to facilitate chemical interpretation of the decisions made by the machine learning method that will be useful to understand and interpret the mass spectra for chemical reactivity.

Highlights

Tandem mass spectrometry (MS/MS) is a powerful analytical tool that is extensively used for the characterization of complex mixtures in many elds, such as proteomics, petroleomics, and drug discovery.[1,2,3,4] Currently, the most commonly used MS/MS technique to obtain structural information for ionized and isolated mixture components is collision-activated dissociation (CAD).[5,6] In these experiments, the analyte ions are accelerated and allowed to collide with an inert gas, such as helium
The major advantage of decision tree models, with analytes represented as an input bit vector of functional groups, is that the resulting ow chart diagram can be interpreted by chemists to gain a deeper understanding of the chemistry resulting in a reaction taking place
We used bootstrapping of several decision tree models to ensure robustness of our model for prospective experimental validations

Summary

Introduction

Tandem mass spectrometry (MS/MS) is a powerful analytical tool that is extensively used for the characterization of complex mixtures in many elds, such as proteomics, petroleomics, and drug discovery.[1,2,3,4] Currently, the most commonly used MS/MS technique to obtain structural information for ionized and isolated mixture components is collision-activated dissociation (CAD).[5,6] In these experiments, the analyte ions are accelerated and allowed to collide with an inert gas, such as helium. Part of the kinetic energy of the ions is converted into their internal energy, resulting in fragmentation. This approach is limited by the fact that isomeric ions o en generate identical fragmentation patterns, making identi cation of compounds via CAD mass spectra unreliable.[4,7] To address this issue, a MS/MS approach based on diagnostic, reliable and predictable gas-phase ion–molecule reactions has been developed.[7,8,9,10,11] This approach can be used to identify speci c functional groups or their combinations in ionized and isolated mixture components to thereby facilitate the differentiation of isomeric ions, o en without the need for reference compounds. The only modi cation to any commercial ion trap or multiquadrupole instrument is the addition of an inlet system for the neutral reagents, which is straightforward.[7,8,12,13]

Objectives

Methods

Results

Conclusion