Interpretation of type 2 diabetes mellitus relevant GC-MS metabolomics fingerprints by using random forests

Huang, Jianhua; Xie, Hualin; Yan, Jun; Cao, Dong‐Sheng; Lü, Hongmei; Xu, Qingsong; Liang, Yi‐Zeng

doi:10.1039/c3ay40379c

Cited by 14 publications

(8 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As an analytical technique, NMR is a powerful approach for both identification and quantification of analytes with a number of important advantages, such as being highly reproducible, high-throughput, non-destructive, and non-biased, not requiring prior chromatographic separation, and, most importantly requiring minimal sample preparation (Bouatra et al 2013 ; Monteiro et al 2013 ; Zhang et al 2013 ; Abuhijleh et al 2009 ; Blindauer et al 1997 ; Di Gangi et al 2014 ). In addition, unlike GC–MS (Bouatra et al 2013 ; Huang et al 2013a ; van der Kloet et al 2012 ; Abbiss et al 2012 ; Kotlowska et al 2011 ; Kuhara et al 2011 ) and certain liquid chromatography–mass spectrometry (LC–MS) methods (Bouatra et al 2013 ; Huang et al 2013 ; Lu et al 2013 ; Norskov et al 2013 ; Boudonck et al 2009 ; Gowda et al 2009 ; Tsutsui et al 2010 ), no chemical derivatization or ionization are necessary. NMR is particularly amenable to detecting polar and uncharged compounds, such as sugars, amines or relatively small, volatile compounds (such as formic acid, formaldehyde, acetone, etc., which are often undetectable by LC–MS methods.…”

Section: Nmr Spectroscopymentioning

confidence: 99%

“…Instead, some key examples are highlighted to illustrate the potential power of the approach. An area of extensive research is the study of cancer biomarkers, including the development of putative biomarkers for bladder cancer (Alberice et al 2013 ; Zhang et al 2012 ), lung cancer (Carrola et al 2011 ), prostate cancer (Duarte and Gil 2012 ; Kumar et al 2014 ), liver cancer (Chen et al 2011 ), colorectal cancer (Wang et al 2010 ), esophageal cancer (Davis et al 2012 ), and kidney cancer (Huang et al 2013c ; Weiss and Kim 2012 ; Kim et al 2011 ; Ganti and Weiss 2011 ). Examples of the metabolites identified as putative biomarkers of kidney cancer include acylcarnitines, quinolinate, gentisate, and 4-hydroxybenzoate.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Standardizing the experimental conditions for using urine in NMR-based metabolomic studies with a particular focus on diagnostic studies: a review

et al. 2014

View full text Add to dashboard Cite

The metabolic composition of human biofluids can provide important diagnostic and prognostic information. Among the biofluids most commonly analyzed in metabolomic studies, urine appears to be particularly useful. It is abundant, readily available, easily stored and can be collected by simple, noninvasive techniques. Moreover, given its chemical complexity, urine is particularly rich in potential disease biomarkers. This makes it an ideal biofluid for detecting or monitoring disease processes. Among the metabolomic tools available for urine analysis, NMR spectroscopy has proven to be particularly well-suited, because the technique is highly reproducible and requires minimal sample handling. As it permits the identification and quantification of a wide range of compounds, independent of their chemical properties, NMR spectroscopy has been frequently used to detect or discover disease fingerprints and biomarkers in urine. Although protocols for NMR data acquisition and processing have been standardized, no consensus on protocols for urine sample selection, collection, storage and preparation in NMR-based metabolomic studies have been developed. This lack of consensus may be leading to spurious biomarkers being reported and may account for a general lack of reproducibility between laboratories. Here, we review a large number of published studies on NMR-based urine metabolic profiling with the aim of identifying key variables that may affect the results of metabolomics studies. From this survey, we identify a number of issues that require either standardization or careful accounting in experimental design and provide some recommendations for urine collection, sample preparation and data acquisition.

show abstract

Section: Nmr Spectroscopymentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Standardizing the experimental conditions for using urine in NMR-based metabolomic studies with a particular focus on diagnostic studies: a review

et al. 2014

View full text Add to dashboard Cite

show abstract

“…Metabolic biomarker discovery is an important aim of metabolomics studies. In model construction, the purpose of variable selection is to find the best combination of variables, which provide the best classification result (Huang et al, 2013). A measure of how each feature contributes to the prediction performance of RF can be calculated in the course of training, and its importance score (VIM) was obtained.…”

Section: Resultsmentioning

confidence: 99%

“…On average, each tree is grown using about 2/3 of the training data, leaving about 1/3 as OOB. The RF algorithm can be stated as follows (Huang et al, 2013):…”

Section: Methodsmentioning

confidence: 99%

NMR Metabolomics and Random Forests Models to Identify Potential Plasma Biomarkers of Blood Stasis Syndrome With Coronary Heart Disease Patients

et al. 2019

View full text Add to dashboard Cite

BackgroundCoronary heart disease (CHD) remains highly prevalent and is one of the largest causes of death worldwide. Blood stasis syndrome (BSS) is the main syndrome associated with CHD. However, the underlying biological basis of BSS with CHD is not yet been fully understood.Materials and MethodsWe proposed a metabolomics method based on 1H-NMR and random forest (RF) models to elucidate the underlying biological basis of BSS with CHD. Firstly, 58 cases of CHD patients, including 27 BSS and 31 phlegm syndrome (PS), and 26 volunteers were recruited from Xiangya Hospital affiliated to Central South University. A 1 mL venous blood sample was collected for NMR analysis. Secondly, principal component analysis (PCA), partial least squares discrimination analysis (PLS-DA) and RF was applied to observe the classification of each group, respectively. Finally, RF and multidimensional scaling (MDS) were utilized to discover the plasma potential biomarkers in CHD patients and CHD–BSS patients.ResultsThe models constructed by RF could visually discriminate BSS from PS in CHD patients. Simultaneously, we obtained 12 characteristic metabolites, including lysine, glutamine, taurine, tyrosine, phenylalanine, histidine, lipid, citrate, choline, lactate, α-glucose, β-glucose related to the CHD patients, and Choline, β-glucose, α-glucose and tyrosine were considered as potential biomarkers of CHD–BSS.ConclusionThe combining of 1H-NMR profiling with RF models was a useful approach to analyze complex metabolomics data (should be deleted). Choline, β-glucose, α-glucose and tyrosine were considered as potential biomarkers of CHD–BSS.

show abstract

“…For example, diabetes patients can be diagnosed as diabetes clinically by some commonly used clinical indicators. However, metabolomics studies by gas chromatography–mass spectrometry have also suggested the difference between diabetes patients and healthy controls by checking some small molecular metabolites . In essence, these two types of data should have certain association.…”

Section: Introductionmentioning

confidence: 99%

Sparse canonical correlation analysis applied to ‐omics studies for integrative analysis and biomarker discovery

Cao

Shao²,

Zeng

et al. 2015

Journal of Chemometrics

Self Cite

View full text Add to dashboard Cite

With the rapid development of new -omics measurement methods, there is an increasing interest in studying the correlation structure between two or more data sets. Multivariate methods such as canonical correlation analysis (CCA) have been proposed to analyze the intrinsic correlation relationship by integrating two data sets. However, because of the high dimensionality of data and the relative scarcity of samples, the ordinary CCA is usually faced with variable selection problems and thereby fails to obtain a satisfactory relationship. Here, we explored the potential of sparse CCA (SCCA) to find the correlative components in two sparse views. SCCA aims at finding sparse projection directions to well extract the correlation between two data sets. We applied this method to one simulation data and one real -omics data to illustrate the performance of SCCA. The results from two studies show that SCCA could effectively find the correlated patterns between two data sets, which are of high importance for understanding the relationship between two underlying chemical or biological processes. The corresponding variable subsets selected by sparse weight vectors can assist in a better interpretation of the chemical or biological process. The integrative analysis from two views by SCCA helps in improving the discriminative ability of classification models for various -omics studies.

show abstract

Interpretation of type 2 diabetes mellitus relevant GC-MS metabolomics fingerprints by using random forests

Cited by 14 publications

References 44 publications

Standardizing the experimental conditions for using urine in NMR-based metabolomic studies with a particular focus on diagnostic studies: a review

Standardizing the experimental conditions for using urine in NMR-based metabolomic studies with a particular focus on diagnostic studies: a review

NMR Metabolomics and Random Forests Models to Identify Potential Plasma Biomarkers of Blood Stasis Syndrome With Coronary Heart Disease Patients

Sparse canonical correlation analysis applied to ‐omics studies for integrative analysis and biomarker discovery

Contact Info

Product

Resources

About