Automated metabolite identification and quantification using J-resolved NMR spectroscopy

We will develop free, publicly accessible software that can be assessed via the web to help measure the amounts of naturally occurring chemicals (called metabolites) in a biological sample. Specifically, the software will analyse data collected using a tool called nuclear magnetic resonance (NMR) spectroscopy, and will determine unambiguously which metabolites are present and will measure their concentrations. This type of study / called metabolomics / has recently grown into a major international field of research following on from the success of genomics and proteomics. Metabolomic studies span from basic research in universities to applied studies in industry, including wide ranging applications such as helping scientists to investigate how herbicides alter the biochemistry of plants; characterising the differences between the metabolism of healthy and diseased patients in a hospital; and investigating how animals in the environment cope with pollution. To date, most metabolomics studies have compared the overall patterns or fingerprints of the metabolites between two or more types of biological samples. Although this can provide important information as to whether overall metabolic differences exist between these samples, it provides very little insight into precisely how these samples differ, in particular how the amount of the individual metabolites vary between the samples. Our proposed methods are an important step forward as they will enable the unambiguous identification and accurate quantification of metabolites. This is important and in fact necessary for several reasons: - It allows new insight into the mechanism of biological processes; e.g., understanding the mechanism of a disease can help to develop new drugs and enable the response of patients to these drugs to be monitored. - It allows changes in the metabolites to be integrated with changes in gene expression (transcriptomics) and protein levels (proteomics) / which is a goal of systems biology / to provide a more complete description of a biological organism. - It allows a biological sample to be truly characterised in terms of its component parts, not just an overall metabolic fingerprint. This in turn allows the NMR data to be compared with measurements obtained using other techniques. The specific objectives of our study are to develop computer software that can automatically analyse metabolite data collected from an NMR spectrometer, specifically from a 2-dimensional (2D) J-resolved NMR dataset. This type of NMR experiment is rapidly gaining in popularity in metabolomics. Since the measurements are displayed in a 2D format, all the hundreds of peaks that arise from the metabolites are spread out significantly, minimising their overlap (which is a major problem in traditional one-dimensional NMR experiments for which unambiguous metabolite identification and quantification is impossible). Our new analysis will produce a list of the metabolites present and their concentrations. Alongside this work we will construct a library of 2D J-resolved NMR datasets of 300 pure metabolites. This important resource is crucial for our analysis method, will be widely valued by the international metabolomics community (see letters from Imperial College London and the University of California-Davis), and will receive the widest possible dissemination as it will be housed in the largest public database of metabolite spectra in the world (see letter from University of Madison-Wisconsin, US). This NMR metabolite library will also be housed at Birmingham, since it forms a critical component of the software package for analysing the 2D NMR data. Therefore we propose to implement a database at Birmingham to store the NMR library. The final outcome will be a user-friendly software package that will be freely available to, and will significantly benefit, the scientific community.

Metabolomics has recently become a versatile tool which is broadly used by industry and academia in the biological sciences. In several recent notable studies the analysis of the small molecule composition of a biological system has provided unprejudiced insight into the metabolic consequences of genetic manipulations, diseases and toxic insult. Although the leading metabolomics technologies have proven relatively successful, one-dimensional (1D) nuclear magnetic resonance (NMR) spectroscopy suffers from severe spectral overlap limiting the ability to identify and quantify metabolites, and accurate quantification of hundreds of metabolites simultaneously is currently unfeasible by mass spectrometry. The full potential of metabolomics cannot be achieved until methods are developed to unambiguously identify and accurately quantify metabolites. Two-dimensional J-resolved (JRES) NMR spectroscopy is rapidly gaining in popularity, and benefits from a dispersion of peaks into a second dimension, improving metabolite specificity and accuracy of quantification. Here we propose to develop a free, web accessible software package for quantitative metabolic profiling based on 2D JRES NMR spectroscopy. This will incorporate a novel 2-step screening/deconvolution algorithm to identify and quantify metabolites in JRES spectra of biological samples, along with an NMR library containing JRES spectra of ca. 300 metabolites. Therefore we also propose to expand our existing BBSRC-funded 1D 1H NMR metabolite library to include 2D JRES spectra at both 500 and 600 MHz. A MySQL database with web accessibility will be constructed to house the spectra of pure metabolites and biological samples. This database will be compliant with the new NMR data standards endorsed by the Metabolomics Standards Initiative. Overall, we anticipate that this software package and associated 2D JRES spectral libraries will become an immediate and important resource for the international metabolomics community.