Curve-Fitting Method for Direct Quantitation of Compounds in Complex Biological Mixtures Using ¹H NMR:  Application in Metabonomic Toxicology Studies

Abstract

A new software tool has been developed that provides automated measurement of signal intensities in NMR spectra of complex mixtures without using data reduction procedures. The algorithm finds best-fit transformations between signals in reference compound spectra and the corresponding signals in analyte spectra. Unlike other algorithms, it is insensitive to variation in chemical shift and can even be used for relative quantitation of compounds whose identities have not yet been established. Additionally, the parameters of the transformation provide information and error metrics that may assist in the streamlining of quality control. The approach presented is general in scope but has been tested by application to peak quantitation in NMR spectra of biofluids. Replicate NMR measurements of solutions of biologically important compounds at various concentrations were made. Further NMR data were collected on urine samples from human, rat, and mouse, which were "spiked" with reference compound solutions at known concentrations. Finally, existing data from an independent toxicology project involving several hundred samples were analyzed, and the consistency of the measurements for metabolites that give multiple NMR signals was assessed. The results of all these tests give confidence that the technique can be used in automated quantitation of compounds in large NMR data sets with minimal operator intervention.

Application of abductive ILP to learning metabolic network inhibition from temporal data

“…The observable data is up/down regulation of metabolites obtained from NMR spectra. The technique which has been used to transform raw time-series data is described in Crockford et al (2005). The up/down regulations of metabolites at different time periods are then encoded as Prolog ground facts 8 .…”

1H NMR metabonomics of plasma lipoprotein subclasses: elucidation of metabolic clustering by self‐organising maps

“…Having said that, it is good to keep in mind that molecular level biochemical understanding is likely to call for focused individual spectral analysis in various applications (65). Therefore, it is good to note that both frequency (66) and time domain (67) data analysis methods are also extensively developing towards fast and automated handling of metabonomic data. Figure 4 shows the U-matrix for the largest SOM (3604 nodes), shown also in Figs.…”

Spin–spin coupling information is crucial for unbiased NMR analysis in metabonomics

“…In order to make the comparisons more reasonable, we report the power of T H for conditions (4), (5), and (6) using F (1) n , F (2) n , and F (3) n (in addition to χ 2 2 ) as the null distribution because, except for the μs, the pairs (1) and (4), (2) and (5), and (3) and (6) have the same parameter values in τ , ρ, and λ. Table 3 shows that T H also has satisfactory power: 0.864, 0.822 and 0.614 for conditions (4), (5), and (6) when F (·) n is conservatively taken as the null distribution.…”

“…However such an assumption is likely violated due to the overlapping peaks, which is further complicated by differences of peak position, line width and baseline correction. Hence bin integration does not necessarily reflect the true changes in spectral areas (Crockford, et al, 2005;Potts, et al, 2001); the concentration information is incoherently evaluated, and is potentially biased.…”