EpiScape: Bioinformatic reconstruction and interactive visualisation of the epigenetic landscape of normal and aberrant hematopoiesis

• Applicant: Florian Halbritter, Ph.D.
• Subject Area: Bioinformatics and Theoretical Biology ; General Genetics and Functional Genome Biology ; Developmental Biology ; Epidemiology and Medical Biometry/Statistics ; Hematology, Oncology
• Term: from 2015 to 2018
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 284000042

With the International Human Epigenome Consortium, a major international effort is underway to establish a comprehensive catalogue of the human epigenome. The results are highly valuable for dissecting the dynamics of cell differentiation and for charting the epigenetic basis of diseases such as cancer. However, these data are difficult to analyze and interpret on a genome-wide scale because of their volume and complexity. Here, I propose a novel approach to addressing the interpretation bottleneck, complementing classical bioinformatician-led analysis with genome-wide exploration by biologists without the need for special bioinformatics training. To this end, I will develop web-based software that makes it straightforward to visually explore enormous data collections -- interactively, in real time, and benefitting from current machine learning and statistical methods behind the scenes. In this way, access to massive-scale epigenome datasets is opened up to basic researchers and biomedical staff, substantially increasing the practical use of the international investments into the data generation projects. I will test-drive the proposed method in a practical application to the stratification of leukemia samples based on epigenome data, with the perspective of developing epigenetic biomarker panels that are relevant for precision medicine.The proposed project comprises three specific aims:AIM 1. AUTOMATED EPIGENOME DATA EXTRACTION AND REPRESENTATION. I will develop and implement a bioinformatics framework for automated extraction of information-rich signatures ("EpiCubes") from collections of epigenome browser tracks. By condensing large quantities of epigenome data into a standardized coordinate system, EpiCubes will effectively provide multi-scale summaries for efficient querying and data integration in real-time.AIM 2. INTERACTIVE VISUALIZATION AND MODELING SOFTWARE. I will develop user-friendly, web-based software ("EpiScape") for real-time visual exploration and modeling of epigenome data from thousands of samples. EpiScape will shift the focus from the locus-centric view of genome browsers towards global patterns across many samples and conditions and it will do so interactively, thus leveraging the interpretative skills of human expert users.AIM 3. EPIGENETIC LANDSCAPE OF NORMAL AND ABERRANT HEMATOPOIESIS. I will use data from BLUEPRINT and the International Human Epigenome Consortium to model the epigenetic landscape of normal hematopoiesis with EpiScape. I will then map thousands of leukemia DNA methylomes on top to investigate aberrant differentiation and leukemia subtypes, as well as potential epigenetic biomarkers for the distinction between subtypes.

• DFG Programme: Research Fellowships
• International Connection: Austria
• Host: Professor Dr. Christoph Bock