A Conversation on Data Mining Strategies in LC-MS Untargeted Metabolomics: Pre-Processing and Pre-Treatment Steps

Abstract

Untargeted metabolomic studies generate information-rich, high-dimensional, and complex datasets that remain challenging to handle and fully exploit. Despite the remarkable progress in the development of tools and algorithms, the “exhaustive” extraction of information from these metabolomic datasets is still a non-trivial undertaking. A conversation on data mining strategies for a maximal information extraction from metabolomic data is needed. Using a liquid chromatography-mass spectrometry (LC-MS)-based untargeted metabolomic dataset, this study explored the influence of collection parameters in the data pre-processing step, scaling and data transformation on the statistical models generated, and feature selection, thereafter. Data obtained in positive mode generated from a LC-MS-based untargeted metabolomic study (sorghum plants responding dynamically to infection by a fungal pathogen) were used. Raw data were pre-processed with MarkerLynxTM software (Waters Corporation, Manchester, UK). Here, two parameters were varied: the intensity threshold (50–100 counts) and the mass tolerance (0.005–0.01 Da). After the pre-processing, the datasets were imported into SIMCA (Umetrics, Umea, Sweden) for more data cleaning and statistical modeling. In addition, different scaling (unit variance, Pareto, etc.) and data transformation (log and power) methods were explored. The results showed that the pre-processing parameters (or algorithms) influence the output dataset with regard to the number of defined features. Furthermore, the study demonstrates that the pre-treatment of data prior to statistical modeling affects the subspace approximation outcome: e.g., the amount of variation in X-data that the model can explain and predict. The pre-processing and pre-treatment steps subsequently influence the number of statistically significant extracted/selected features (variables). Thus, as informed by the results, to maximize the value of untargeted metabolomic data, understanding of the data structures and exploration of different algorithms and methods (at different steps of the data analysis pipeline) might be the best trade-off, currently, and possibly an epistemological imperative.

Mass spectrometry in untargeted liquid chromatography/mass spectrometry metabolomics: Electrospray ionisation parameters and global coverage of the metabolome

“…The cleaned data are less 'noisy' and contain mainly defined features. The latter refers to the 'relevant' ions/true signals (each with a unique m/z and Rt) that are extracted from the raw data, capturing as much usable information as possible . Furthermore, for semantic clarification, the term 'defined features' (specifically in this study) means recorded true MS signals, relevant ionised species (extracted from processing and data scrutiny steps – clean data matrices), which most likely represent 'detected' metabolites.…”

“…Before performing PCA, the data were mean‐centered (to put all variables on equal footing) and Pareto‐scaled to adjust for measurement errors . Nonlinear iterative partial least‐squares (NIPALS) algorithm (with a correction factor of 3.0) was the methodology used to handle the missing values, with a default threshold of 50%. A seven‐fold cross‐validation (CV) procedure was applied as a tuning method to build the models; and only the components producing an increase in the prediction ability of the model ( R1 significant components) were retained.…”

“…Another contributing factor is the inherent complexity of metabolomes due to the physicochemical diversity, instability, and rapid turnover of the cellular metabolites . To achieve a wide coverage of the metabolome, despite the current limitations, untargeted (data‐driven) metabolomic strategies have been heuristically advised and applied, profiling the largest number of compounds possible and therefore (potentially) elucidating previously unexplored biochemical processes …”

Food Metabolomics: A New Frontier in Food Analysis and its Application to Understanding Fermented Foods

“…MVDA tools commonly used in food metabolomics studies include artificial neural networks (ANN), principal component analysis (PCA), orthogonal projection to latent structures-discriminant analysis (OPLS-DA), partial least square discriminant analysis (PLS-DA), principal component regression (PCR), hierarchical cluster analysis (HCA), canonical correlation analysis (CCA) and others [16,38,43]. Detailed strategies, algorithms and explanation on these MVDA techniques have been described in detail elsewhere [24,39,[43][44][45][46].…”

“…Handling these huge data would require an automated software for quantification and identification [24]. Pretreatment basically involves alignment, normalization, compound identification, centering, transformation, scaling, removing baseline artefacts and peak picking [16,24,38,39], in order to convert the raw data set into a form that can be utilized for subsequent analysis. Succeeding analysis of the cleaned data in food metabolomics studies are majorly done using different chemometric tools, to provide a description and understanding of the variations and/or similarities in the metabolites.…”