Abstract
In high-throughput scenarios, a large number of similar NMR spectra from a single study need to be analyzed. The complexity of the data required the use of spectral databases to characterize resonances of interest and extract quantitative information. The recently reported CRAFT (Complete Reduction to Amplitude Frequency Table) technique, based on a Bayesian analysis approach, converts a time-domain FID to a frequency-amplitude table in a robust, automated, and time-efficient fashion. We report the application of the CRAFT technique to the extraction of quantitative information from the NMR spectra of complex mixtures – the targeted analysis of spent media from mammalian cell cultures and the untargeted profiling of soy supplement extracts. CRAFT, in a significantly automated fashion, converts the raw NMR spectra into a data-mining-friendly spreadsheet format with high fidelity and accuracy. The reported examples clearly demonstrate that the automated, time-domain data reduction by the CRAFT technique gives comparable results to traditional approaches. Moreover, the approach described herein allows for iterative adjustment of the post- CRAFT NMR parametric filters of the tabular data (such as linewidth and/or amplitude and/or frequency window thresholds) for reexamination. Such iterative parametric filters in conjunction with statistical analysis/guidance have the potential opportunity to develop analyte fingerprint databases for subsequent sample screening and library comparisons. Thus, this technique potentially allows for the development of rapid screening methods, both targeted and untargeted, to be implemented easily, and be employed effectively in high throughput environments.
Keywords: Automated NMR analysis, bayesian, fermentanomics, food sciences, mixture analysis, quality control, quantitative NMR.