These are preliminary reports that have not been peer-reviewed. They should not be regarded as conclusive, guide clinical practice/health-related behavior, or be reported in news media as established information. For more information, please see our FAQs.
2 files

Graph Based Machine Learning Interprets Diagnostic Isomer-Selective Ion-Molecule Reactions in Tandem Mass Spectrometry

submitted on 28.12.2019, 02:16 and posted on 31.12.2019, 20:55 by Jonathan A Fine, Judy Kuan-Yu Liu, Armen Beck, Kawthar Alzarieni, Xin Ma, Victoria Boulos, Hilkka Kenttämaa, Gaurav Chopra
Diagnostic ion-molecule reactions using tandem mass spectrometry can differentiate between isomeric compounds unlike a popular collision-activated dissociation methodology for the identification of previously unknown mixtures. Selected neutral reagents, such as 2-methoxypropene (MOP) are introduced into an ion trap mass spectrometer and react with protonated analytes to produce product ions diagnostic of the functional groups present in the analyte. However, the interpretation and understanding of specific reactions are challenging and time-consuming for chemical characterization. Here, we introduce a first bootstrapped decision tree model trained on 36 known ion-molecule reactions with MOP using graph-based connectivity of analyte’s functional groups as input. A Cohen Kappa statistic of 0.72 was achieved, suggesting substantial inter-model reliability on limited training data. Prospective diagnostic product predictions were made and validated for 14 previously unpublished analytes . Chemical reactivity flowcharts were introduced to understand the decisions made by the machine learning method that will be useful for chemists.


Integrated Data Science Institute Award

Department of Chemistry Start-up Funds at Purdue University

Purdue University Center for Cancer Research, NIH grant P30 CA023168


Email Address of Submitting Author


Purdue University


United States

ORCID For Submitting Author


Declaration of Conflict of Interest


Version Notes