[Stellenangebote] PhD/Master project in ML and computational biology

[Markus.Mittnenzweig at mdc-berlin.de]

Hi everyone,

we have exciting PhD or Master projects to work on machine learning models of gene regulation during embryonic development, together with the Robson lab at MDC-BIMSB! For the PhD position, please apply by March 3 through the current MDC PhD call.

👉 https://mdc-berlin.de/phd#PhDjob

Please get in touch for more details! Informal enquiries are encouraged.

Kind regards,
Markus


---
Example project: Machine learning the enhancer grammar controlling embryonic development

The Mittnenzweig and Robson labs seek to examine how gene expression controls cell-fate decisions in embryonic development and human disease. The Mittnenzweig group focus on developing computational methods to model cell-fate decisions from single cell data. The Robson lab implements genomics methods in mice and organoids to profile how chromatin chromatin and its 3D structure implement these decisions.

Mammalian organogenesis lays the blueprint for an embryo’s major organs and is highly conserved. To achieve this, our genomes coordinate precise changes to gene expression that guide cells along defined cell-fates. However, we still do not fully understand the molecular machinery that encodes this regulation. In particular, transcription factors bind specific regulatory DNA sequences to instruct where and when a gene should activate. It remains unclear what “grammar” of DNA sequence permits transcription factor binding or which events this triggers on chromatin to activate genes.

In this joint project, the Mittnenzweig and Robson labs seek to address this by leveraging recent advances in single-cell multi-omics and machine leaning. Novel multi-omics methods in the Robson lab can now map protein binding and regulatory element activity simultaneously with gene expression in a single cell. Similarly, new machine learning tools can then use these data to extract the DNA sequences that underlie these regulatory activities. Combining these tools will thus reveal the steps genomes take to activate cells and control cell fate.

The project will consist of two main steps that examine the developing mouse limb, our in vivo model of embryogenesis. First, you will use single-cell multiome (RNA + ATAC) measurements to construct a quantitative model of the cellular differentiation trajectories that form the limb. Second, you incorporate single cell measurements of chromatin state and structure into the model and collectively map the epigenetic dynamics of each trajectory. Finally, you will learn machine learning models from these data, thereby predicting which epigenetic changes define cell-fate decisions. Any such model will subsequently be validated and tested by perturbing specific transcription factors or characterizing additional epigenetic modalities.

[cid:image001.png at 01DA52DF.5316D1C0] [cid:image002.jpg at 01DA52DF.5316D1C0]

Figure 1. Single-cell multiome profiling of early mammalian development. UMAP of ~60.000 single-nuclei multiome profiles (RNA + ATAC) of late mouse gastrulation and early organogenesis between E7.5 and E8.75. Image taken from Argelaguet et al1 (left). Microscopy image of the developing limb bud (right)





---

Markus Mittnenzweig

Computational and Developmental Bioloy Lab



Max Delbrück Center

Berlin Institute for Medical Systems Biology (MDC-BIMSB)

Hannoversche Straße 28

10115 Berlin



Phone: +49 30 9406-1339

markus.mittnenzweig at mdc-berlin.de<mailto:markus.mittnenzweig at mdc-berlin.de>

www.mdc-berlin.de/mittnenzweig<http://www.mdc-berlin.de/mittnenzweig>

www.twitter.com/mittnenzweig<http://www.twitter.com/mittnenzweig>