Computational Epigenomics

dna_methylation
Christoph Bock, Max Planck Institute for Informatics, DNA methylation, CC BY-SA 3.0

Machine Learning for High Throughput Bio-Medical Data

We adapt machine learning tools to promote the understanding of epigenetic control mechanisms for gene expression and their function in development and disease.

What is Epigenetics?

200 highly specialized cell types in the human body (muscle cells or nerve cells) contain identical copies of DNA. How are cellular differences achieved?

Parts of the DNA are condensed such that they can’t be targeted by the cellular transcription mechanism. Dynamic compaction of chromatin is mediated by epigenetic mechanisms including chemical modifications of the DNA (e.g. DNA methylation), and of histone tails and nucleosome repositioning.

Genome wide epigenomic snapshots can be generated by using Next Generation Sequencing.

How can Machine Learning help?

Data output grows exponentially, leading to a bottleneck in data analysis, integration and visualization. In the absence of precise biological models, ML methods are particularly suited to untangle complex relationships from large, high dimensional, noisy datasets.