ANYL 207 |
| Given the volume and complexity of spectroscopic data that can be obtained from metabonomics studies, it is necessary to invoke pattern recognition (PR) methods to ensure optimum retrieval of information. The PR process has two often overlapping and undistinguishable stages: discovery and classification. PR is traditionally applied to metabolic data for classification; yet the discovery, exploratory part of PR can give us a great insight into the metabolic interaction and pathways. Often the performance of PR is impeded by experimental and instrument-induced variations which in general obscure the process of pattern discovery. Frequency shifts are the dominant source of such unwanted variations in spectral datasets. A technique for corrections of frequency shifts will be presented, as well as application of Principal Component Analysis for detection, identification and removal, if needed, of sources of variation in metabonomic datasets. Several PR approaches for increasing the sensitivity and resolution of the NMR experiments will be discussed. |
|
Development and Applications of Metabonomic/Metabolomic Methods of Analysis
1:30 PM-4:30 PM, Monday, 11 September 2006 Moscone Center -- Room 124, Oral
Division of Analytical Chemistry |