Verarbeitung und funktionelle Analyse von molekularen Profilen aus Microarray-Experimenten
Felix Kokocinski
Data management and functional analysis of molecular profiles from microarray experiments The application of high-throughput technologies like matrixCGH and expression profiling with microarrays in biomedical research, allows for the highly parallel analysis of clinical samples. In order to conduct these kinds of experiments and cope with the amount of data produced, specialized software solutions are needed. For the management of a clone collection as the starting point for all experiments, the CloneBase system was developed. It comprises of a database with internet access pages and automated data-update functions. It permits the archiving, annotation and information retrieval of all clones. The nucleic acid fragments of selected clones are used for the production of microarrays with the help of laboratory robots. To operate and control the complex production process, a laboratory information management system (LIMS) was developed. QuickLIMS can process the barcode-identified microtiter plates containing the fragments, using a specified protocol. The user interface is generated dynamically from the workflow, which itself is defined in a flexible manner. The system allows a maximum of process automation and error tracking for microarray production. The integration of bio-molecular data from diverse sources like public databases or clinical knowledge (annotation data) is a key challenge in the process of the analysis of these high-throughput experiments. The Flexible Annotation and Correlation Tool (FACT) was developed as a flexible framework for the explorative meta-analysis of genomic, proteomic or other experiments. Heterogeneous experimental data and annotational sources can be integrated and diverse analytical algorithms can be applied with the goal of finding distinct patterns and illuminating inherent characteristics within the data. The results from different experimental systems, e.g. information on genomic changes from matrixCGH analysis and expression profiles generated by cDNA microarrays can be brought into correlation using dedicated program modules. This is achieved through the abstraction from specific data types into a general data scheme and by the flexible storage of data source definitions. The FACT system is easily extendible to cover additional sources and include new methods. Finally, the application AutoPrime automates the process of selecting primer sequences for RQ-PCR as used for the verification of the microarray results. It links a sequence database with a primer design program, making the selection faster and less error-prone. With the applications introduced here, a database and analysis system was established which facilitates the experimental process of biomedical high-throughput research. It allowed the successful analysis of the pathomechanisms of different brain tumors (menignioma and medulloblastoma) and of hematopoetic topics (differentiation of the cell line HL60, acute myeloic leukemia). In particular the development and progression of non-melanoma skin cancer was comprehensively analyzed with microarrays containing murin fragments. This allowed the identification of new candidate genes for human skin cancer, among others members of the epidermal differentiation complex (1q21.3).

[Full text]     [back]