Computational Methods for the Discovery of Metabolic Markers of Complex Traits

Abstract

Metabolomics uses quantitative analyses of metabolites from tissues or bodily fluids to acquire a functional readout of the physiological state. Complex diseases arise from the influence of multiple factors, such as genetics, environment and lifestyle. Since genes, RNAs and proteins converge onto the terminal downstream metabolome, metabolomics datasets offer a rich source of information in a complex and convoluted presentation. Thus, powerful computational methods capable of deciphering the effects of many upstream influences have become increasingly necessary. In this review, the workflow of metabolic marker discovery is outlined from metabolite extraction to model interpretation and validation. Additionally, current metabolomics research in various complex disease areas is examined to identify gaps and trends in the use of several statistical and computational algorithms. Then, we highlight and discuss three advanced machine-learning algorithms, specifically ensemble learning, artificial neural networks, and genetic programming, that are currently less visible, but are budding with high potential for utility in metabolomics research. With an upward trend in the use of highly-accurate, multivariate models in the metabolomics literature, diagnostic biomarker panels of complex diseases are more recently achieving accuracies approaching or exceeding traditional diagnostic procedures. This review aims to provide an overview of computational methods in metabolomics and promote the use of up-to-date machine-learning and computational methods by metabolomics researchers.

Discrimination of rosé wines using shotgun metabolomics with a genetic algorithm and MS ion intensity ratios

“…Despite this very important selection, a similar reasoning led to the possibility of 5112 ratios of this 72 ion peaks, which lead to 3.3 × 10 30 possible subsets of size 1 to 10 and to 4.6 × 10 20 years of computation for a comprehensive search of the best subset. In this context, usual analysis workflows would fail and powerful heuristic search algorithms are required 14 . We chose a genetic algorithm which has often been used in feature selection contexts [15][16][17] including metabolomics biomarkers studies 18,19 .…”

“…In order to assess the adequacy of our approach, we compared it to a very usual method for biomarkers analysis in metabolomics, Random Forests (RF) 14,[20][21][22] . Obviously, there is no embedded method in RF to allow any selection based on ratios.…”

Use of Integrative Interactomics for Improvement of Farm Animal Health and Welfare: An Example with Fescue Toxicosis

“…Further, recent comparisons between pipelines leading to amplicon sequence variants (ASVs) and OTUs seem to indicate some differences in specificity between the two, with different OTU-based pipelines producing some non-overlapping spurious OTUs [ 133 ]. For metabolomics, differences in analytical platforms, computational methods, and database matching algorithms for metabolomics could also result in metabolic feature variability [ 134 , 135 , 136 , 137 ]. Finally, while still a less visible, yet budding field, multi-‘omics integration methodologies, such as xMWAS [ 21 ], can vary, and their continual development and improvement will increase consistency.…”

Deep metabolome: Applications of deep learning in metabolomics

“…number of layers and their connections) and structures with less than two hidden layers are called shallow ANNs, while more complicated architectures are found in the larger class of Deep Neural Network (DNN) which can be more expressive and efficient than their simpler ANN variants [20] . For reviews of introductory ANN and DL methodology, which is outside the scope of this review, we refer readers to other articles containing historical and methodological perspectives [21] , [22] .…”