The General Explanation Method with NMR Spectroscopy Enables the Identification of Metabolite Profiles Specific for Normal and Tumor Cell Lines

Abstract

Machine learning models in metabolomics, despite their great prediction accuracy, are still not widely adopted owing to the lack of an efficient explanation for their predictions. In this study, we propose the use of the general explanation method to explain the predictions of a machine learning model to gain detailed insight into metabolic differences between biological systems. The method was tested on a dataset of 1H NMR spectra acquired on normal lung and mesothelial cell lines and their tumor counterparts. Initially, the random forests and artificial neural network models were applied to the dataset, and excellent prediction accuracy was achieved. The predictions of the models were explained with the general explanation method, which enabled identification of discriminating metabolic concentration differences between individual cell lines and enabled the construction of their specific metabolic concentration profiles. This intuitive and robust method holds great promise for in‐depth understanding of the mechanisms that underline phenotypes as well as for biomarker discovery in complex diseases.

Children gut microbiota exhibits a different composition and metabolic profile after in vitro exposure to Clostridioides difficile and increases its sporulation

“…The area under each binned region was integrated and the integrated binned regions were normalized to the 'constant sum' equal to 100. 2D NMR spectra were processed and analyzed using VNMRJ (Agilent-Varian) and Sparky (UCSF) software as previously described (Pecnik et al, 2018). Random Forests analysis was performed using MATLAB's TreeBagger algorithm (MATLAB Statistics and Machine Learning Toolbox, v.R2016b) with 30 bagged classification trees.…”

“…The perfect prediction accuracy demonstrated the ability of the random forests model to successfully learn of 1 H NMR binned regions, which discriminated the three clusters of supernatants. Next, the general explanation method was applied to accurately highlight the most important binned regions used for predictions by the random forests model (Pecnik et al, 2018). NMR signals in the most important binned regions of 1 H NMR spectra were assigned and used to identify nine metabolites with the use of 2D 13 C HSQC, 13 C HMBC and TOCSY spectra.…”