A comparative evaluation of the generalised predictive ability of eight machine learning algorithms across ten clinical metabolomics data sets for binary classification DOI Creative Commons
Kevin Mendez, Stacey N. Reinke, David Broadhurst

et al.

Metabolomics, Journal Year: 2019, Volume and Issue: 15(12)

Published: Nov. 15, 2019

Abstract Introduction Metabolomics is increasingly being used in the clinical setting for disease diagnosis, prognosis and risk prediction. Machine learning algorithms are particularly important construction of multivariate metabolite Historically, partial least squares (PLS) regression has been gold standard binary classification. Nonlinear machine methods such as random forests (RF), kernel support vector machines (SVM) artificial neural networks (ANN) may be more suited to modelling possible nonlinear covariance, thus provide better predictive models. Objectives We hypothesise that classification using metabolomics data, non-linear will superior generalised ability when compared linear alternatives, particular with current PLS discriminant analysis. Methods general performance eight archetypal across ten publicly available data sets. The were implemented Python programming language. All code results have made Jupyter notebooks. Results There was only marginal improvement SVM ANN over all RF comparatively poor. use out-of-bag bootstrap confidence intervals provided a measure uncertainty model prediction quality observed bigger influence on than choice. Conclusion size set, choice metric, had greater algorithm.

Language: Английский

MetaboAnalyst 4.0: towards more transparent and integrative metabolomics analysis DOI Creative Commons

Jasmine Chong,

Othman Soufan,

Carin Li

et al.

Nucleic Acids Research, Journal Year: 2018, Volume and Issue: 46(W1), P. W486 - W494

Published: April 13, 2018

We present a new update to MetaboAnalyst (version 4.0) for comprehensive metabolomic data analysis, interpretation, and integration with other omics data. Since the last major in 2015, has continued evolve based on user feedback technological advancements field. For this year's update, four key features have been added 4.0, including: (1) real-time R command tracking display coupled release of companion MetaboAnalystR package; (2) MS Peaks Pathways module prediction pathway activity from untargeted mass spectral using mummichog algorithm; (3) Biomarker Meta-analysis robust biomarker identification through combination multiple datasets (4) Network Explorer integrative analysis metabolomics, metagenomics, and/or transcriptomics The interface 4.0 reengineered provide more modern look feel, as well give space flexibility introduce functions. underlying knowledgebases (compound libraries, metabolite sets, metabolic pathways) also updated latest Human Metabolome Database (HMDB). A Docker image is available facilitate download local installation MetaboAnalyst. freely at http://metaboanalyst.ca.

Language: Английский

Citations

3369
metaX: a flexible and comprehensive software for processing metabolomics data DOI Creative Commons
Bo Wen, Zhanlong Mei,

Chunwei Zeng

et al.

BMC Bioinformatics, Journal Year: 2017, Volume and Issue: 18(1)

Published: March 21, 2017

Non-targeted metabolomics based on mass spectrometry enables high-throughput profiling of the metabolites in a biological sample. The large amount data generated from requires intensive computational processing for annotation spectra and identification metabolites. Computational analysis tools that are fully integrated with multiple functions easily operated by users who lack extensive knowledge programing needed this research field.We herein developed an R package, metaX, is capable end-to-end through set interchangeable modules. Specifically, metaX provides several functions, such as peak picking annotation, quality assessment, missing value imputation, normalization, univariate multivariate statistics, power sample size estimation, receiver operating characteristic analysis, biomarker selection, pathway correlation network metabolite identification. In addition, offers web-based interface ( http://metax.genomics.cn ) assessment normalization method evaluation, it generates HTML-based report visualized interface. utilities were demonstrated published dataset scale. software available operation either graphical user (GUI) or form command line functions. package example reports at http://metax.genomics.cn/ .The pipeline platform-independent easy to use spectrometry.

Language: Английский

Citations

672
Tau PTM Profiles Identify Patient Heterogeneity and Stages of Alzheimer’s Disease DOI Creative Commons

Hendrik Wesseling,

Waltraud Mair,

Mukesh Kumar

et al.

Cell, Journal Year: 2020, Volume and Issue: 183(6), P. 1699 - 1713.e13

Published: Nov. 13, 2020

To elucidate the role of Tau isoforms and post-translational modification (PTM) stoichiometry in Alzheimer's disease (AD), we generated a high-resolution quantitative proteomics map 95 PTMs on multiple isolated from postmortem human tissue 49 AD 42 control subjects. Although PTM maps reveal heterogeneity across subjects, subset display high occupancy frequency for AD, suggesting importance disease. Unsupervised analyses indicate that occur an ordered manner, leading to aggregation. The processive addition minimal set associated with seeding activity was further defined by analysis size-fractionated Tau. summarize, features protein critical intervention at different stages are identified, including enrichment 0N 4R isoforms, underrepresentation C terminus, increase negative charge proline-rich region (PRR), decrease positive microtubule binding domain (MBD).

Language: Английский

Citations

519
Missing Value Imputation Approach for Mass Spectrometry-based Metabolomics Data DOI Creative Commons
Runmin Wei, Jingye Wang, Mingming Su

et al.

Scientific Reports, Journal Year: 2018, Volume and Issue: 8(1)

Published: Jan. 8, 2018

Missing values exist widely in mass-spectrometry (MS) based metabolomics data. Various methods have been applied for handling missing values, but the selection can significantly affect following data analyses. Typically, there are three types of not at random (MNAR), (MAR), and completely (MCAR). Our study comprehensively compared eight imputation (zero, half minimum (HM), mean, median, forest (RF), singular value decomposition (SVD), k-nearest neighbors (kNN), quantile regression left-censored (QRILC)) different using four datasets. Normalized root mean squared error (NRMSE) NRMSE-based sum ranks (SOR) were to evaluate accuracy. Principal component analysis (PCA)/partial least squares (PLS)-Procrustes used overall sample distribution. Student's t-test followed by correlation was conducted effects on univariate statistics. findings demonstrated that RF performed best MCAR/MAR QRILC favored one MNAR. Finally, we proposed a comprehensive strategy developed public-accessible web-tool application ( https://metabolomics.cc.hawaii.edu/software/MetImp/ ).

Language: Английский

Citations

515
Metformin Enhances Autophagy and Normalizes Mitochondrial Function to Alleviate Aging-Associated Inflammation DOI Creative Commons
Leena P. Bharath, Madhur Agrawal,

Grace McCambridge

et al.

Cell Metabolism, Journal Year: 2020, Volume and Issue: 32(1), P. 44 - 55.e6

Published: May 12, 2020

Language: Английский

Citations

452
Quality assurance procedures for mass spectrometry untargeted metabolomics. a review DOI
Danuta Dudzik,

Cecilia Barbas-Bernardos,

Antonia Garcı́a

et al.

Journal of Pharmaceutical and Biomedical Analysis, Journal Year: 2017, Volume and Issue: 147, P. 149 - 173

Published: Aug. 5, 2017

Language: Английский

Citations

306
The gut metabolite indole-3 propionate promotes nerve regeneration and repair DOI

Elisabeth Serger,

L. Gutiérrez, Jessica Chadwick

et al.

Nature, Journal Year: 2022, Volume and Issue: 607(7919), P. 585 - 592

Published: June 22, 2022

Language: Английский

Citations

210
Navigating freely-available software tools for metabolomics analysis DOI
Rachel Spicer, Reza M. Salek, Pablo Moreno

et al.

Metabolomics, Journal Year: 2017, Volume and Issue: 13(9)

Published: Aug. 9, 2017

Language: Английский

Citations

200
Metabolomics of exhaled breath in critically ill COVID-19 patients: A pilot study DOI Creative Commons
Stanislas Grassin‐Delyle, Camille Roquencourt,

Pierre Moine

et al.

EBioMedicine, Journal Year: 2020, Volume and Issue: 63, P. 103154 - 103154

Published: Dec. 4, 2020

Early diagnosis of coronavirus disease 2019 (COVID-19) is the utmost importance but remains challenging. The objective current study was to characterize exhaled breath from mechanically ventilated adults with COVID-19. In this prospective observational study, we used real-time, online, proton transfer reaction time-of-flight mass spectrometry perform a metabolomic analysis expired air undergoing invasive mechanical ventilation in intensive care unit due severe COVID-19 or non-COVID-19 acute respiratory distress syndrome (ARDS). Between March 25th and June 25th, 2020, included 40 patients ARDS, whom 28 had proven multivariate analysis, identified characteristic breathprint for We could differentiate between ARDS accuracy 93% (sensitivity: 90%, specificity: 94%, area under receiver operating curve: 0·94-0·98, after cross-validation). four most prominent volatile compounds were methylpent-2-enal, 2,4-octadiene 1-chloroheptane, nonanal. non-invasive detection nonanal may identify funded by Agence Nationale de la Recherche (SoftwAiR, ANR-18-CE45-0017 RHU4 RECORDS, Programme d'Investissements d'Avenir, ANR-18-RHUS-0004), Région Île France (SESAME 2016), Fondation Foch.

Language: Английский

Citations

194
Identification of key taste components in loquat using widely targeted metabolomics DOI

Shicheng Zou,

Jincheng Wu, Muhammad Qasim Shahid

et al.

Food Chemistry, Journal Year: 2020, Volume and Issue: 323, P. 126822 - 126822

Published: April 18, 2020

Language: Английский

Citations

193