Professor Inge Jonassen and part of his team from the CBU (Computational Bioinformatics Unit, Department of Informatics, UiB) are working on the development and application of bioinformatics methods for analysing data descending from high-throughput measurement technologies applied to cancer samples.
The primary focus of the group, in collaboration with the Akslen group, is the development and application of computational deconvolution methods for decomposing transcriptome data from samples composed of a combination of tumor cells and the surrounding and supporting microenvironment. The research aims to decompose computationally the signal into that originating from the tumor cells and those originating from other tissues/cell types in the sample. This will be an enabling step towards studying the interactions between tumor cells and the environment and integrating them into the research along the continuum from diagnosis to treatment and outcome.
The group’s projects
The group’s first objective was to analyse several public benchmark transcriptome datasets (microarray and RNA-Seq) with a variety of econvolution methods, test their performance and find out which of the mathematical assumptions reflect biological reality to a sufficient degree and can lead to more robust results. On second level, they developed a new computational method that addresses several computational challenges and they show extensively the performance of the group’s method relative to other state-of-the-art methods (manuscript under preparation) based on the benchmark data. The group’s proposed approach on breast cancer data was further tested and biologically consistent results were found. In this framework, they also developed SelGenes, a tool for selecting marker genes (i.e. genes highly specific for a tissue/cell type) for the cell types included in heterogeneous samples (published in master thesis). The performance of the group’s approach was tested both on the benchmark data and on cancer data and biologically consistent results were found.
Another focus of the group is the development and application of systems biology integrative approaches. In collaboration with the Akslen group, they applied a subpathway enrichment analysis approach to reveal mechanisms that change between tumor samples with high and low Nestin expression, associated with the basal-like phenotype in breast cancer (the results are part of the analysis of a submitted manuscript). Moreover, in collaboration with the Biosignal Lab (Professor Anastasios Bezerianos, Deptartment of Medical Physics, School of Medicine, University of Patras, Greece), the group developed a time-varying method for microRNA-mediated subpathway enrichment analysis. The tool was tested on interferon-gamma (IFN-g) stimulated melanoma cells.
One manuscript (submitted), in collaboration with the Akslen group. One manuscript (in preparation), in collaboration with the Akslen group. Master thesis, “SelGenes: a tool for selecting marker genes in heterogeneous samples”, Kristian Samdal. Conference abstract: Konstantina Dimitrakopoulou, Elisabeth Wik, Lars Akslen, Inge Jonassen. Gene expression deconvolution in complex tissues via particle swarm optimization. EMBL Cancer Genomics, 1 - Nov 4 2015, Heidelberg, Germany. Conference abstract: Konstantina Dimitrakopoulou, Elisabeth Wik, Lars Akslen, Inge Jonassen. Deconvolution of transcriptome data from heterogeneous tissue samples. 15th European Conference on Computational Biology, Sept 3-7 2016, Hague, Netherlands. Journal publication: Vrahatis AG, Dimitrakopoulou K, Balomenos P, Tsakalidis AK, Bezerianos A. CHRONOS: a time-varying method for microRNA-mediated subpathway enrichment analysis.Bioinformatics. 2016 Mar 15;32(6):884-92.
Plans for the future
In the longer perspective, the Jonassen group intends to apply network-based approaches on the deconvoluted expression data to explore the interactions involved in different tumor types and their microenvironments. Furthermore, they intend to integrate other omics data like DNA methylation, copy number variation and protein expression, and also to explore utlisation of singlecell omics data, to improve the group’s comprehension of the mechanisms underlying tumor development and treatment response. Furthermore, the group intends to be more involved in applied work including data sets generated within the center.ose D, Lin K, Sheldon T, Jonassen I. 2008. Prediction of protein structure from ideal forms. Proteins: Struct, Funct, and Bioinfo. 70, 1610-1619.