Gene set based integrated data analysis reveals phenotypic differences in a brain cancer model.

A key challenge in the data analysis of biological high-throughput experiments is to handle the often low number of samples in the experiments compared to the number of biomolecules that are simultaneously measured. Combining experimental data using independent technologies to illuminate the same biological trends, as well as complementing each other in a larger perspective, is one natural way to overcome this challenge. In this work we investigated if integrating proteomics and transcriptomics data from a brain cancer animal model using gene set based analysis methodology, could enhance the biological interpretation of the data relative to more traditional analysis of the two datasets individually. The brain cancer model used is based on serial passaging of transplanted human brain tumor material (glioblastoma--GBM) through several generations in rats. These serial transplantations lead over time to genotypic and phenotypic changes in the tumors and represent a medically relevant model with a rare access to samples and where consequent analyses of individual datasets have revealed relatively few significant findings on their own. We found that the integrated analysis both performed better in terms of significance measure of its findings compared to individual analyses, as well as providing independent verification of the individual results. Thus a better context for overall biological interpretation of the data can be achieved.