Freie Universität Berlin
FB Mathematik + Informatik
Institut für Mathematik
Phone: 030 - 838 51445|
Fax: 030 - 838 75412
|Homepage Conrad:||Homepage Conrad|
|Homepage Group:||Homepage Group|
Proteomics is a relatively young discipline dealing with the complicated protein mixture in each part of the human or animal body. Opposed to e.g. genomics, the players in this area of molecular biology and biochemistry are single molecules like proteins. Their concentration in body fluids (such as blood, urine or eye and brain fluid) often span many orders of magnitude during a day, depending on the varying conditions the human body is exposed to.
During the last years, several new emerging proteomic methods show promise in non-invasive screening of the body fluids mentioned above. The characterization of peptides and proteins in serum and plasma by mass spectrometry (MS) is one of the promising strategies for biomarker discovery. The computer-assisted detection of molecules provides a snapshot of thousands of peptides, protein fragments and proteins. This new analytical technology has the potential to identify disease-associated proteomic patterns in blood serum.
Clinical proteomics is an exciting new sub-discipline of proteomics that involves bedside application of proteomic technologies. Recently introduced special mass spectrometry based screening methods, e.g. surface enhanced and matrix assisted laser desorption/ionization-time of flight-mass spectrometry (SELDI-TOF-MS and MALDI-TOF-MS, respectively) have fostered the development of a new and potentially revolutionary technology and approach for early disease detection, surveillance, and monitoring: proteomic pattern diagnostics. Following and enhancing this approach, a newly developed standardized magnetic bead based pre-analytical process followed by a high throughput MALDI-TOF mass spectrometry (Baumann et al.) generates a standardized proteomic fingerprint of a given body fluid, such as serum. In this proteome profile, biomarkers are sought that allow to discriminate healthy and diseased individuals.
The aim of our main project in this area is to develop statistical algorithms capable of identifying disease (e.g. cancer) specific features (or sets of features) of proteome profiles using meticulously classified patient samples. The data archives we are working with typically contain profiles of over 1000 individuals including meta data (such as age, gender, and disease status). Normally, this datasets have been assembled by our collaborators. Special emphasis is laid on the applicability in clinical areas, i.e. algorithms are sought for which are simple, fast and reliable, with particular respect to sensitivity and specificity sufficient for clinical requirements.
To be able to cope with the vast amount of data (several Terra Bytes) new algorithms must be developed and existing technologies enhanced. The main (computer science) areas we are dealing with here are database technologies, workflow based distributed (PC cluster) and parallel computing techniques and new methods for visualization of these huge datasets. This research is complemented by the development of novel statistical and combinatorial techniques for pattern recognition, cluster identification, dimension reduction and outlier detection (see MATHEON-Project A10).
Ultimately, a web-based proteomics workbench is being developed that enables researchers from different disciplines (e.g. biology, medicine, bioinformatics, computer science, mathematics) to use the newly developed algorithms on their own data or on existing data in our repositories. Therefore, not only a new MS database will be created but also available high-throughput data analysing techniques will be available to process this data.