Analytic Properties of Statistical Total Correlation Spectroscopy Based Information Recovery in ¹H NMR Metabolic Data Sets

Abstract

Structural assignment of resonances is an important problem in NMR spectroscopy and Statistical Total Correlation Spectroscopy (STOCSY) is a useful tool aiding this process for small molecules in complex mixture analysis and metabolic profiling studies. STOCSY delivers intramolecular information (delineating structural connectivity) and in metabolism studies can generate information on pathway related correlations. To understand further the behavior of STOCSY for structural assignment, we analyze the statistical distribution of structural and non-structural correlations from 1050 1 H NMR spectra of normal rat urine samples. We find that the distributions of structural/non-structural correlations are significantly different (p<10 -112 ). From the area under the curve of the receiver operating characteristic (ROC AUC) we show that structural correlations exceed non-structural correlations with probability AUC=0.98. Through a bootstrap resampling approach, we demonstrate that sample size has a surprisingly small effect (e.g. AUC=0.97 for a sample size of 50). We identify specific signatures in the correlation maps resulting from small matrixderived variations in peak positions but find that their effect on discrimination of structural and non-structural correlations is negligible for most metabolites. A correlation threshold of r>0.89 is required to assign two peaks to the same metabolite with high probability (positive predictive value, PPV=0.9), while sensitivity and specificity are equal at 93% for r=0.22. To assess the wider applicability of our results, we analyze 1 H NMR spectra of urine from rats treated with 115 model toxins or physiological stressors. Across the data sets, we find that the thresholds required to obtain PPV=0.9 are not significantly different and the degree of overlap between the structural and nonstructural distributions is always small (median AUC=0.97).The STOCSY method is effective for structural characterization under diverse biological conditions and sample sizes provided the degree of correlation resulting from non-structural associations (e.g. from non-stationary processes) is small. This study validates the use of the STOCSY approach in the routine assignment of signals in NMR metabolic profiling studies and provides practical benchmarks against which researchers can interpret the results of a STOCSY analysis.

Coadministration of the Campylobacter jejuni N-Glycan-Based Vaccine with Probiotics Improves Vaccine Performance in Broiler Chickens

“…STOCSY using the resonance at 5.0 ppm as a driver peak also yielded strong positive correlations (0.91 and 0.82) with resonances at 2.9 and 2.7 ppm and a negative correlation (Ϫ0.59) with the bacterial load (Fig. S3A), suggesting that the features at 5.0, 2.9, and 2.7 ppm are in the same molecule (40). Two-dimensional (2D) NMR experiments (heteronuclear single quantum coherence total correlation spectroscopy [HSQC-TOCSY] and TOCSY) confirmed the STOCSY results showing that the three statistically correlated features are in the same molecule (Fig.…”

13C NMR Metabolomics: Applications at Natural Abundance

“…We calculated 2D statistical correlation maps for 13 C– 13 C and 1 H– 1 H homonuclear (STOCSY: 10,11 Figure 1C) and 1 H– 13 C heteronuclear spectra (SHY: 12−14 Figure 1D). In 1D versions of STOCSY and SHY, correlations and covariance are determined for a specific “driver” peak of interest.…”

“…We analyzed the 13 C and 1 H 1D data sets with 13 C– 13 C statistical total correlation spectroscopy (STOCSY) 10,11 and 13 C– 1 H statistical heterospectroscopy (SHY) 12−14 to compile correlated peak lists that were then used in database searches for compound identification. We developed and validated the workflow using synthetic mixtures of 20 common metabolites, defined the limits of detection using current technology, and compared the peak lists generated using both 13 C and 1 H data to those generated using 1 H data alone.…”

MetaboHunter: an automatic approach for identification of metabolites from 1H-NMR spectra of complex mixtures

“…Various approaches were described in previous publications, including: (i) binning approaches [8,9] where a spectrum is typically divided into equally or variable sized bins and the intensities in each bin are qualified and quantified via integration techniques; (ii) curve fitting without reference library support, where de-convoluting highly overlapped linearly mixed individual metabolite spectra is achieved via various methodologies ranging from Bayesian decompositions [10,11] and least squares-based non-negative matrix factorization [12] to shape fitting techniques [13-17]; (iii) curve fitting with reference library support, where least squares strategies [18-20], Bayesian model selection [21], and genetic algorithms [22,23] are employed, and (iv) direct comparison methods that calculate the overlap of known peaks with peaks from query spectra [22]. More comprehensive descriptions of methods and practical aspects of applied metabolomics are described in a number of recent publications [24-27].…”