Research Group Seifert
The working group analyses data from various analytical methods, which can be regarded as fingerprints of the samples examined. We are working at the interface of method development/validation of chemometric and bioinformatic approaches, mainly performed with simulated data, and their practical applications in various research areas. Of particular interest are omics data sets, which are high-dimensional data derived from the analysis of an entire cellular analyte group, thereby representing its entirety. We are also very interested in the analysis of various spectroscopic data, specifically surface-enhanced Raman scattering, e.g. to study the influence of drugs in living cells. Our analyses aim at the characterization of the samples beyond their black box classification. This means that important variables are selected and interactions of different variables are analyzed to assess the properties of the sample in as much detail as possible.
The objectives of the data analysis and the various areas of the group's work are presented below. (Graphics: freisinn.net)
Different analytical techniques are applied to generate a specific fingerprint of the chemical and biological identity of food. We exploit this fingerprint for classification and characterization by the application of various chemometric and bioinformatic approaches.
In the framework of the Cluster of Excellence “Understanding Written Artefacts”, chemometric and bioinformatic approaches are developed and applied to contribute to the characterization of the biological identity and historical background of written artefacts. A main focus is currently on the analysis of genetic and infrared spectroscopic data from palm leaf manuscripts.
Random Forest (RF) is a machine learning method that consists of a multitude of individual decision trees. RF based approaches are applied, optimized and validated in the Seifert group.
Surface-enhanced Raman Scattering (SERS)
In SERS experiments Raman signals are enhaned by several orders of magnitude. As a result, local information from the surrounding of the nanoparticles is obtained. When analyzing complex biological samples, such as cells, data is obtained that consists of superimposed signals from different molecules and is therefore difficult to evaluate. The working group is developing methods to extract as much information as possible from this data.