Automatic reduction of NMR spectroscopic data for statistical and pattern recognition classification of samples.

A general method of automatically reducing NMR spectra to provide numerical descriptors of samples has been developed and investigated. These descriptors can be used as input to pattern recognition or multivariate algorithms for sample classification. The methods have been tested using 600 MHz one-dimensional 1H NMR spectra of biofluids which are complex mixtures. The approach is, in principle, applicable to multidimensional and heteronuclear NMR spectra and to other types of liquid samples such as oils and foodstuffs as well as to situations such as 1H or 31P NMR in vivo and solid state NMR in drug formulation analysis. The method relies upon apportioning the information in the spectra to individual contiguous segments and allowing specified regions of the spectra to be omitted. Three approaches, based on the number of peaks, the summed peak heights and the summed peak areas respectively in each segment, have been tested. The effect of segment width and overlap and the effects of manipulation of the NMR spectra have been evaluated in terms of the classification of the samples using principal components analysis. A simple method of generating NMR based spectral descriptors for object classification is thus proposed.

Automatic data reduction and pattern recognition methods for analysis of 1H nuclear magnetic resonance spectra of human urine from normal and pathological states.

Multivariate data analysis techniques have been used to compare 600-MHz 1H nuclear magnetic resonance (NMR) spectra of urine obtained from patients with inborn errors of metabolism (IEM) and urine obtained from healthy subjects. These spectra are very complex; each contains many thousands of resonances with a high dynamic range. A consistent method of reducing this wealth of data to manageable proportions is presented as a two-stage process. Computer-based spectral descriptors are automatically generated and then reduced to two-dimensional maps for visualization of clustering. Data-scaling methodology has been developed to achieve complete separation between spectra from control adults and those from adult patients with independently diagnosed IEM. The methods were refined by relating IEM samples to the mean of the control samples and applying supervised learning techniques to identify descriptors contributing to class separation. This approach allowed separation of the various classes of IEM and achieved optimal separation of patients with cystinuria from those with oxalic aciduria; the principal metabolites responsible for this separation were determined as lysine and glyoxalate. The methods developed were then extended by application to the more subtle problem of classifying urine collected from healthy subjects under different physiological conditions (i.e., pre- and post-exercise and in different stages of hydration) where, unlike the IEM case, any underlying biochemical differences were not known at the outset. Fluid-loaded and fluid-deprived samples could be partially separated as well as fluid-deprived and fluid-restored samples. Partial classification of samples on the basis of subject was also observed. Therefore, intersubject differences were liable to obscure the separation by physiological state. However, by relating each sample to a mean of the normal daily urine samples for the same person and applying a form of "range scaling" to exclude data which contributed least to class separation, improved classification of the hydration states resulted, from which it was possible to deduce those biochemical substances which were altered. These novel techniques for the data reduction and classification of NMR spectra make comprehensive use of all of the NMR spectral information and have clear potential to assist in clinical diagnosis.

Nuclear magnetic resonance spectroscopy and pattern recognition analysis of the biochemical processes associated with the progression of and recovery from nephrotoxic lesions in the rat induced by mercury(II) chloride and 2-bromoethanamine.

Nephrotoxic lesions were induced in Fischer 344 rats using HgCl2, a proximal tubular toxin, and 2-bromoethanamine (BEA), a medullary toxin. Biochemical effects of these toxins on urinary composition were observed by high resolution 1H NMR spectroscopy over 9 days after dosing. The onset of, progression of, and recovery from the induced toxic lesions were also followed histopathologically and related to the perturbed urinary biochemistry. Urinary concentrations of 20 endogenous substances were measured simultaneously by NMR at eight time points, to provide a time-related 20-dimensional description of the urinary biochemistry for each rat. Principal components analysis and nonlinear mapping were used to reduce the biochemical parameter spaces for each rat to two or three dimensions for display and classification purposes. An investigation of alternative data-presentation methods was made, and taking interanimal means of the map coordinates at each time point yielded a novel type of metabolic trajectory diagram with which the biochemical abnormalities associated with the HgCl2 and BEA lesions could be related to the progression and recovery phases of the toxic lesions. The time-course trajectories showed characteristically different paths for each toxin. These trajectories allowed the time points at which there were maximum metabolic differences to be determined and provided the visualization of net movements of the treatment group populations in time in relation to interanimal variation. Control animal urine samples subjected to this analysis showed simple clustering, with no evidence of metabolic trajectory. The trajectory for BEA showed different routes for onset of and recovery from toxicity, whereas for HgCl2 the outward trajectory (onset) mapped a space similar to the inward trajectory (recovery phase). This suggests that the NMR-detectable biochemical abnormalities after mercury toxicity mainly reflect the proportions of functional cells lining the nephron, whereas the biochemical abnormalities associated with renal medullary insult probably relate to functional integrity. An examination has been made for those metabolites that are most responsible for defining the trajectories, i.e., the discrimination of renal cortical and medullary toxicity from each other and from controls. These discriminatory metabolites (using paired t test, p < 0.001) included valine, taurine, trimethylamine N-oxide, and glucose for HgCl2 and acetate, methylamine, dimethylamine, lactate, and creatine for BEA, whereas citrate, succinate, N-acetyl resonances from as yet unidentified metabolites, hippurate, alanine, and 2-oxoglutarate played an important role in defining the biochemically perturbed trajectory of both toxins.(ABSTRACT TRUNCATED AT 400 WORDS)

An automatic data reduction and transfer method to aid pattern recognition analysis and classification of NMR spectra.

A method of automatically generating reduced NMR data and transferring it between computers is proposed. These data can then be used as descriptors for input to non-parametric statistical routines for classification of the samples.

A quantitative structure-activity relationship approach to the minimization of albumin binding.

The binding of 2,6-disubstituted xanthones to human serum albumin (HSA) has been investigated using an ultrafiltration technique. A set of 26 compounds was chosen for study using a selection procedure aimed at minimizing the interparameter correlations, while ensuring that the physicochemical properties covered the maximum possible range of values. The magnitude of binding has been expressed as the compound concentration required to produce a specified bound concentration, in preference to equilibrium constants and number of albumin binding sites. Albumin binding was found to have a nonlinear dependence on the octanol-water partition coefficient (log P) and has been rationalized in terms of a simple binding model.

Pattern recognition display methods for the analysis of computed molecular properties.

Pattern recognition methods, particularly the 'unsupervised learning' techniques, are well suited for the preliminary analysis of the large data sets produced by computer chemistry. The use of linear and non-linear display methods for such exploratory analysis are exemplified with the aid of two data sets of biologically active molecules. Advantages and disadvantages of these techniques are discussed.

Physicochemical characteristics of non-electrolytes and their uptake by Brugia pahangi and Dipetalonema viteae.

The uptake of a diverse set of 14C-labelled non-electrolytes by Brugia pahangi and Dipetalonema viteae was measured relative to the free diffusion of tritiated water. Inulin was used as a non-absorbable surface marker to account for non-electrolyte adherent to the surface of the parasite which had not crossed the cuticle. B. pahangi and D. viteae took up the non-electrolytes to a similar degree; a comparison of tissue uptake indices gave a correlation coefficient of 0.99. Worm uptake could not be described by non-electrolyte octanol/aqueous partition coefficients alone. However, greater success was achieved using further descriptors and pattern recognition techniques for data analysis. The whole molecule descriptors log P, molar refraction, melting point, dipole moment and CNDO total energy were obtained from computer chemistry and the literature. Using a linear learning machine to relate uptake to these 5 physicochemical descriptors it was possible to successfully classify non-electrolytes as high or low uptake. Multivariate regression analysis of uptake versus these 5 parameters gave a correlation coefficient of 0.77. However, this was not statistically significant and therefore could not be used for quantitative predictions of substance uptake by worms. This illustrates the value of 'pattern recognition' techniques such as the linear learning machine. Using such 'pattern recognition' methods on a chemically related set of compounds it is anticipated that predictions of uptake can be achieved and improved upon. Such predictions could then be used in drug design.