Genomic data integration & gene expression analysis of single cells

Today microarray gene expression profiling has become a routine high throughput method in modern molecular biology laboratories. Gene expression data is used to characterize and distinguish tissues, cells or tumors. However, the regulation and interaction of genes is still not fully understood. In this context combination of gene expression data with other per gene measurements like transcription factor binding or response to a certain stimulus can improve the understanding of the underlying regulatory mechanisms. Further, this approach enables us to categorize gene expression profiles according to the additional information like the presence of a transcription factor. In this thesis a novel data integration algorithm has been developed and proven to work properly in the desired data setting. Guided clustering complements existing standard approaches and even outperforms them. The application of guided clustering to a lymphoma cancer data set discovered a novel independent factor for survival and supported the hypothesis of Toll-like receptor signaling mediating BCL6 induced gene regulation. Another trend in the field of gene expression profiling is the analysis of single cells. The employment of microarray technology for single cell analysis is rather new, but applications range from the analysis of rare cell populations, like disseminated tumor cells, to fertility medicine or embryogenesis, where single cells are of major importance. This thesis evaluates the prospects of single cell gene expression analysis using the Operon GeneChip platform. Further, this platform is applied to a data set from developmental biology. This analysis investigates two questions: (i) How is cellular pluripotency established within the zygote? (ii) Is there an asymmetric mRNA distribution present in the early mammalian development, like it is observed in other species like Drosophila or Xenopus? In summary, transcriptomic prepatterning during the first two mitotic divisions is either to subtle or not-existent. Hence, if there exists maternal prepatterning it is more likely to be mediated post translationally. However, transcriptomic differences were identified between spindle and oocyte as well as between Pb2 and zygote profiles