Integrative data analysis

Acquiring high-dimensional global molecular data sets for DNA sequence, associated chromatin features as well as the transcriptome and proteome has become a routine task and provides a rich information source for genome functions and their deregulation in cancer. To exploit these data it is essential to develop integrative mathematical models from which key regulatory steps or aberrant regulation in tumors can be identified. Furthermore, the underlying mechanisms need to be identified to understand how genome related activities are controlled by chromatin features and epigenetic signaling.

We use various modeling approaches to determine the relationship between chromatin states across the genome, to calculate the activity of transcription factors, and to describe chromatin state transitions. For instance, we employ correlation analysis of deep sequencing data to quantitate the co-localization between different states (A) and to identify gene regulatory networks consisting of co-regulated genes (B). Furthermore, we build mechanistic models for gene activation (C) and compare them to the induction kinetics of a transcriptional reporter that is recorded using time-lapse microscopy.

 

Network models. (A) Network representation of the genome-wide colocalization of different chromatin features. (B) Regulatory network for the activity of transcription factors and their target genes. (C) Kinetic model for the transition between an active and an inactive chromatin state.

Key references

Molitor J, Mallm JP, Rippe K & Erdel F (2017). Retrieving chromatin patterns from deep sequencing data with correlation functions. Biophys J 112, 473-490. doi: 10.1016/j.bpj.2017.01.001 |Abstract | Reprint (10.3 MB) | Software | Article metrics.