The Theis-lab is searching for 2 PhDs/Postdocs in computational biology:

 

(1) ‘Generative modeling for data integration and state prediction in single-cell transcriptomics’ 

Background: 

Deep Learning (DL) is currently experiencing an unprecedented speed of development. This development is facilitated by large, curated imaging datasets which are continuously growing. Thus, state-of-the-art methods in DL are typically developed for imaging data. In medical research, molecular features spaces (omics measurements) are often assayed and these are structured differently to pixel-based feature space. A research focus of the ML group is adapting DL approaches built for imaging applications to account for the peculiarities of omics data, in particular single-cell transcriptomics. Moreover, we work on the interpretability of networks and on integrating prior knowledge, in collaborations with experimental partners from leading biomedical labs around the world. No background in biology is required for this position, but your work will have a direct impact on biomedical research via collaborations with experimental groups.

Goals: You will

  • adapt generative models (variational autoencoders, generative adversarial networks) to statistical models describing single-cell expression 
  • use the GAN for data integration and state prediction and compare with existing approaches
  • include biological prior knowledge from known regulatory network modules into the model (such as transcription factor binding sites at promoters, protein-protein interactions etc.)
  • extend the model to use graph convolutions on the gene space.
  • improve and tailor your model or existing methods from the group (such as scGen) to further omics data modalities.

Further reading:

Scanpy – Single-Cell Analysis in Python

 

(2) ‘Deep learning in spatial transcriptomics and proteomics’

Background:

It is possible to measure various features (proteins or mRNA) of cells in their spatial context in a tissue. This implies that one can integrate the cell type with its position in a tissue and with meaningful covariates, such as disease. Computational tools to integrate spatial context are still scarce as the experimental techniques to perform these measurements are relatively new: To date, most architectures proposed in this field rely on densely connected layers as the feature space is difficult to structure - this will change with spatial data. You do not require a background in Biology for this project but your work will directly impact biomedical research and will be in the context of collaborations with experimental groups.

Goals: You will

  • use and adapt AI methods to describe and explore feature spaces both for spatial transcriptomics and proteomics: This involves image segmentation and the usage of coarse grained spatial coordinates with convolutional image processing network architectures.
  • adapt current deep learning network structures to also exploit spatial information: This involves using convolutions in generative models (such as conditional VAEs) supervised settings and standard unsupervised heterogeneity exploration. 
  • work on model interpretation (-> Interpretable Machine Learning): Interpret spatial context models inferred as neural networks.

Further reading:

Scanpy – Single-Cell Analysis in Python

 Reconstructing cell cycle and disease progression using deep learning, Eulenberg, P. et al., Nature Communications (2017)

Prospective identification of hematopoietic lineage choice by deep learning, Buggenthin, F. et al., Nature Methods (2017)

Spatial transcriptomics coming of age, Burgess, D. J., Nature Reviews Genetics (2019)

 Interpretability of deep learning models: A survey of results, Chakraborty, S. et al., IEEE (2017)

Explaining Explanations: An Overview of Interpretability of Machine Learning, Leilani H. et al., in press

 

Your qualification:

  • M. Sc. / PhD in Computer Science / Physics / Math or related field
  • Applied experience with machine/deep learning methods 
  • Overview over latest machine/deep learning research
  • Strong programming skills in python 
  • Ability to handle multiple projects in a dynamic environment
  • Interest in discussing and exploring potential project ideas with collaboration partners
  • Strong interest in medical / biological questions 

Application:

Please send your electronic application (in English) in a single PDF file – including cover letter, CV including a complete list of publications, statement of research interests, and contact details of at least two references to

We are looking forward to receiving your comprehensive online application. Please contact the ICB recruiting team via Email:

 

 

We use cookies to improve your experience on our Website. We need cookies to continuously improve the services, to enable certain features and when embedding services or content of third parties, such as video player. By using our website, you agree to the use of cookies. We use different types of cookies. You can personalize your cookie settings here:

Show detail settings
Please find more information in our privacy statement.

There you may also change your settings later.