Recent studies suggest members of the FGF family as possible targets for new pharmaceuticals in human hepatocellular carcinoma (HCC), a cancer type with low chances of survival due to high resistance against drugs. In this study we analysed the interaction of the FGF family in HCC with an analysis pipeline programmed mainly in R. Data was obtained from the TCGA/GDC database. Besides the pre-processing steps in this pipeline, we also implemented DeMixT, which is a deconvolution algorithm performing an in-silico microdissection on expression data. Microdissection in-vitro is a necessary step to get refined samples for sequencing in cancer, but it is also expensive and time-consuming. Computational approaches, on the other side, are relatively cheap and provide good results. Here we calculated gene expression profiles where no infiltration of stromal and immune cells occurred. We performed differential gene expression analysis, gene set enrichment analysis, gene regulatory network analysis and survival analysis on the given and on the deconvolved data. In both the convoluted and the deconvolved data, the expression of FGF12 and FGF13 is upregulated and the expression of FGF2 is downregulated. To get deeper insight into the resulting gene sets of the gene set enrichment analysis with focus on the members of the FGF family, we created mutual information networks of these gene sets using Aracne-AP. Comparing the deconvolved normal and the deconvolved tumour compartment showed changes of the interaction between FGFs and all other genes. Furthermore, results of the survival analysis of both the convoluted and the deconvolved data overlap on FGF18 and FGFR3. According to these, low chances of survival correlate with low expression of FGF18 and high expression of FGFR3.
