Fröhlich Lab | Scientific Machine Learning for Signal Transduction

A 2024 Crick PhD project with Fabian Fröhlich. This application is open until 12:00 noon on 19 March 2024.
Deadline for applications has passed.

Key information

Applications closed
19 March 2024, 12:00 GMT
Information
Posted 19 February 2024

Research topics

Computational & Systems Biology Signalling & Oncogenes Imaging Biochemistry & Proteomics
Background texture taken from the lab imagery.

A 2024 Crick PhD project with Fabian Fröhlich. 

Project background and description

Signal Transduction allows cells to respond to extracellular cues. Which and how strongly distinct signalling pathways are activated in individual cells varies from cell to cell and depends on their molecular makeup. The impact of the molecular makeup's effect on signalling can be examined through two perspectives: top-down and bottom-up. From the top-down view, molecular makeup is summarized as cell states that indirectly affects signal transduction. Conversely, the bottom-up approach focuses on how the levels of specific proteins directly modifies signal transmission.

In the Fröhlich lab, we adopt a scientific machine learning strategy that integrates mathematical modelling with machine learning to explore the variability in signal transduction across cells. This approach allows us to combine the advantages of both top-down and bottom-up views. Potential PhD projects in our lab include:

  • Deep Mechanistic Models (described in the project video), which blends representation learning with mathematical modelling to predict signalling responses to perturbations from baseline quantifications of the molecular makeup of cells. We are currently focusing on the prediction of signalling responses (measured by mass cytometry) to perturbations with growth factors and kinase inhibitors based on baseline omics data and are interested in extending this to imaging data modalities.
  • Neural Coarse Graining uses neural networks to infer emergent dynamical laws from complex mathematical models of protein-protein interactions. We apply these models to build simple models of signalling pathways that account for the direct influence of protein levels on signal transduction. Moreover, we use neural coarse graining to design synthetic, multicomponent signalling cascade from scratch.
  • Construction of mechanistic models consolidating knowledge graphs, pathway databases and structural information to construct multi-pathway models of signal transduction. Models will be trained on proteomic and phosphoproteomic data.

The specific project will be tailored to the expertise and preferences of the individual candidate, ensuring a meaningful alignment with their background and interests.

Candidate background

Candidates from diverse backgrounds, including but not limited to Systems Biology, Natural Sciences, Bioinformatics, Computer Science, Data Science, Mathematics and Physics, are encouraged to apply. A quantitative mindset, coding experience (Python/Julia/C or similar language) and an interest in solving biological problems is expected.

References

1.         Fröhlich, F., Gerosa, L., Muhlich, J. and Sorger, P.K. (2023)

            Mechanistic model of MAPK signaling reveals how allostery and rewiring contribute to drug resistance.

            Molecular Systems Biology 19: e10988. PubMed abstract

2.         Fröhlich, F., Kessler, T., Weindl, D., Shadrin, A., Schmiester, L., Hache, H., . . . Hasenauer, J. (2018)

            Efficient parameter estimation enables the prediction of drug response using a mechanistic pan-cancer pathway model.

            Cell Systems 7: 567-579.e566. PubMed abstract