Summary
Our abilities to predict and engineer complex biological systems are in their infancy. In the context of gene regulation, we cannot design artificial promoters with specificity to arbitrary cell states, and we cannot arbitrarily trans- and de-differentiate somatic cells, although such abilities would be of high biotechnological and biomedical value. To achieve these ambitious goals, we require quantitative, predictive models of gene regulatory elements (GREs) and gene regulatory networks (GRNs), respectively. Here I propose that the combination of deep learning and single-cell genetic screens is ideally suited to obtain such models, and, in particular, the amounts of highly informative data required for their training. Working with an ex vivo model of hematopoietic stem cell differentiation, we will first screen the activity of hundreds of thousands of synthetic GREs throughout the hematopoietic differentiation landscape. Thereby, we will obtain models of cell type specific GRE activity that can predict new, synthetic GREs with activity in any cell state of interest. Second, we will screen hundreds of thousands of combinatorial GRN perturbations and their effect on cell state. Thereby, we will derive models of GRNs that can predict combinatorial perturbation strategies to achieve arbitrary de- or trans-differentiation events. In sum, work on this project will yield a quantitative model of gene regulation in hematopoiesis at various scales of complexity while introducing a novel, AI-guided concept for biological engineering.
Unfold all
/
Fold all
More information & hyperlinks
| Web resources: | https://cordis.europa.eu/project/id/101041399 |
| Start date: | 01-09-2022 |
| End date: | 31-08-2027 |
| Total budget - Public funding: | 1 499 653,00 Euro - 1 499 653,00 Euro |
Cordis data
Original description
Our abilities to predict and engineer complex biological systems are in their infancy. In the context of gene regulation, we cannot design artificial promoters with specificity to arbitrary cell states, and we cannot arbitrarily trans- and de-differentiate somatic cells, although such abilities would be of high biotechnological and biomedical value. To achieve these ambitious goals, we require quantitative, predictive models of gene regulatory elements (GREs) and gene regulatory networks (GRNs), respectively. Here I propose that the combination of deep learning and single-cell genetic screens is ideally suited to obtain such models, and, in particular, the amounts of highly informative data required for their training. Working with an ex vivo model of hematopoietic stem cell differentiation, we will first screen the activity of hundreds of thousands of synthetic GREs throughout the hematopoietic differentiation landscape. Thereby, we will obtain models of cell type specific GRE activity that can predict new, synthetic GREs with activity in any cell state of interest. Second, we will screen hundreds of thousands of combinatorial GRN perturbations and their effect on cell state. Thereby, we will derive models of GRNs that can predict combinatorial perturbation strategies to achieve arbitrary de- or trans-differentiation events. In sum, work on this project will yield a quantitative model of gene regulation in hematopoiesis at various scales of complexity while introducing a novel, AI-guided concept for biological engineering.Status
SIGNEDCall topic
ERC-2021-STGUpdate Date
09-02-2023
Images
No images available.
Geographical location(s)
