A tutorial review: Metabolomics and partial least squares-discriminant analysis – a marriage of convenience or a shotgun wedding

Gromski, Piotr S.; Muhamadali, Howbeer; Ellis, David I.; Xu, Yun; Correa, Elon; Turner, Michael L.; Goodacre, Royston

doi:10.1016/j.aca.2015.02.012

Cited by 713 publications

(523 citation statements)

References 147 publications

Supporting

Mentioning

514

Contrasting

Unclassified

Order By: Relevance

“…Discriminant analyses of principal components does not require assumptions on a population genetic model (e.g., linkage equilibrium of markers) in contrast to programs like STRUCTURE (Pritchard, Stephens, & Donnelly, 2000) so DAPC has been widely adopted in recent population genetic studies (Buchalski et al., 2016; Cahill & Levinton, 2016; Grünwald & Goss, 2011). Its use for metabolic study is recent, but its efficacy in discriminating different biologically meaningful chemotype classes has been demonstrated and favored over other discriminant analysis methods under certain circumstances (Gromski et al., 2015; Scheitz et al., 2013). …”

Section: Methodsmentioning

confidence: 99%

Discovering variation of secondary metabolite diversity and its relationship with disease resistance inCornus floridaL.

Pais

Xiang

2018

Ecology and Evolution

View full text Add to dashboard Cite

Understanding intraspecific relationships between genetic and functional diversity is a major goal in the field of evolutionary biology and is important for conserving biodiversity. Linking intraspecific molecular patterns of plants to ecological pressures and trait variation remains difficult due to environment‐driven plasticity. Next‐generation sequencing, untargeted liquid chromatography–mass spectrometry (LC‐MS) profiling, and interdisciplinary approaches integrating population genomics, metabolomics, and community ecology permit novel strategies to tackle this problem. We analyzed six natural populations of the disease‐threatened Cornus florida L. from distinct ecological regions using genotype‐by‐sequencing markers and LC‐MS‐based untargeted metabolite profiling. We tested the hypothesis that higher genetic diversity in C. florida yielded higher chemical diversity and less disease susceptibility (screening hypothesis), and we also determined whether genetically similar subpopulations were similar in chemical composition. Most importantly, we identified metabolites that were associated with candidate loci or were predictive biomarkers of healthy or diseased plants after controlling for environment. Subpopulation clustering patterns based on genetic or chemical distances were largely congruent. While differences in genetic diversity were small among subpopulations, we did observe notable similarities in patterns between subpopulation averages of rarefied‐allelic and chemical richness. More specifically, we found that the most abundant compound of a correlated group of putative terpenoid glycosides and derivatives was correlated with tree health when considering chemodiversity. Random forest biomarker and genomewide association tests suggested that this putative iridoid glucoside and other closely associated chemical features were correlated to SNPs under selection.

show abstract

Section: Methodsmentioning

confidence: 99%

Discovering variation of secondary metabolite diversity and its relationship with disease resistance inCornus floridaL.

Pais

Xiang

2018

Ecology and Evolution

View full text Add to dashboard Cite

show abstract

“…A large number of signals could be studied in the discrimination of classes considering the Variable-Importance-in-Projection-(VIP) which was set at a minimum value of 2. Generally, VIP set at 1 can be considered important in a given model 56 . With regard to the reproducibility of the method, this could be considered good with variation coefficients (CV%) below 20% for the metabolites identified.…”

Section: Resultsmentioning

confidence: 99%

Direct infussion Electrospray Mass Spectrometry as a new non-invasive tool for serum metabolomics in induced-stress subjects

et al. 2015

View full text Add to dashboard Cite

-Background and Objectives: Nanotechnology is becoming a tool for the study of changes in the metabolome of patients in different states of disease. Analytical techniques such as Electrospray Mass Spectrometry, allow to find biomarkers by determination of metabolites. Nowadays, there is not an objective analytical approach for diagnosis of stress. Thus, the objectives of this pilot work are: Describing the development of a fast, direct and non-invasive analytical protocol, applied for the first time, to study the metabolomic profile of patient s different states through a disease. Testing the protocol in a pilot sample with non-stressed and stress-induced subjects.Methods: High resolution direct infusion electrospray mass spectrometry has been used to analyse the metabolome of blood samples (0.3 ml) from six subjects.Results: Data prove a clear discrimination between non-stressed and stressed states in the metabolome. Data showed different predominant metabolites in both states. Results allow objective characterization of the state of the patient.Conclusions: Although this is a pilot study, the method was successful in discriminating different metabolites in non-stressed and stress-induced subjects.

show abstract

“…In [7], authors show that PLS-DA outperforms other approaches in terms of feature selection and classification. In a more detailed study [8], authors compare different variable selection approaches such as LDA, PLS-DA, SVM-Recursive Feature Elimination (RFE), RF (with accuracy and gini), for identifying the best suited method for analyzing metabolomic data and classifying the Grampositive bacteria Bacillus.…”

Section: State Of the Artmentioning

confidence: 99%

“…Metabolomic data usually contain highly correlated features, leading to some problems when using RF for example [7]. Filter methods allow to select "good features", such as the "coefficient of correlation" (Cor) or the "mutual information" (MI) measures.…”

Section: The Reduction Of Dimensionalitymentioning

confidence: 99%

A Hybrid Knowledge Discovery Approach for Mining Predictive Biomarkers in Metabolomic Data

Grissa

Comte²,

Pujos‐Guillot³

et al. 2016

Machine Learning and Knowledge Discovery in Databases

View full text Add to dashboard Cite

Abstract. The analysis of complex and massive biological data issued from metabolomic analytical platforms is a challenge of high importance. The analyzed datasets are constituted of a limited set of individuals and a large set of features where predictive biomarkers of clinical outcomes should be mined. Accordingly, in this paper, we propose a new hybrid knowledge discovery approach for discovering meaningful predictive biological patterns. This hybrid approach combines numerical classifiers such as SVM, Random Forests (RF) and ANOVA, with a symbolic method, namely Formal Concept Analysis (FCA). The related experiments show how we can discover among the best potential predictive biomarkers of metabolic diseases thanks to specific combinations of classifiers mainly involving RF and ANOVA. The visualization of predictive biomarkers is based on heatmaps while FCA is mainly used for visualization and interpretation purposes, complementing the computational power of numerical methods.

show abstract

A tutorial review: Metabolomics and partial least squares-discriminant analysis – a marriage of convenience or a shotgun wedding

Cited by 713 publications

References 147 publications

Discovering variation of secondary metabolite diversity and its relationship with disease resistance inCornus floridaL.

Discovering variation of secondary metabolite diversity and its relationship with disease resistance inCornus floridaL.

Direct infussion Electrospray Mass Spectrometry as a new non-invasive tool for serum metabolomics in induced-stress subjects

A Hybrid Knowledge Discovery Approach for Mining Predictive Biomarkers in Metabolomic Data

Contact Info

Product

Resources

About