| Introducing Spectra Classifier |
|
Mass Frontier can be used for the classification of spectra according to physical or chemical properties (point of origin, toxicity, aromaticity etc.). For spectra that cannot be found in a library, classification can involve identification of substructure types or compound classes (structure elucidation) in order to establish and confirm structural proposals. Classification can also be useful in cases when only structurally related compounds need to be retrieved from a complex chromatogram (metabolite research). The Spectra Classifier module allows the user to retrieve and organize spectra, which can then be submitted to Principal Component Analysis (PCA), Self-Organizing Maps (SOM) and Fuzzy Clustering. Spectra are organized into groups that can have their own names and graphic representations. When spectra are organized into groups it is easier to distinguish between the classes of spectra that should arise from the submitted groups of spectra after the classification process. In classification analysis it is important to distinguish between two terms: groups and classes. It does not make sense to speak about classes prior to the classification process. Classes can only be assigned after the classification method (PCA, Fuzzy Clustering or SOM) clearly shows the occurrence of clusters or regions that consist of spectra with the desired properties. Prior to classification, we can only allocate spectra to groups according to user-selected criteria, i.e. a priori information. Often the objective of the classification process is to obtain classes that closely resemble groups of spectra submitted to classification Spectra Classifier topics: See also: |
