Computational Models and Algorithms for Predicting Cell Reprogramming Determinants with High Efficiency and High Fidelity – AlgoReCell

Cell reprogramming consists in acting on specific genes, so-called Reprogramming Determinants (RDs), in order to trigger a cell de- or trans-differentiation. The prediction in-silico of such determinants is a major challenge, notably for regenerative medicine. Recent works have demonstrated with experimental validations that some RDs can be identified from the topology and the discrete Boolean dynamics of gene regulatory networks.

The goal of AlgoReCell is to design a computational framework for the prediction of reprogramming strategies driven by three main challenges:
(1) the systematic and complete identification of RDs according to a discrete model of gene networks;
(2) the assessment of RD-dependent reprogramming efficiency and fidelity;
(3) the selection of best combination of RDs candidates taking into account the uncertainties in the model, and the potential state heterogeneity of the cell population.
The resulting framework will be evaluated experimentally for the reprogramming of adipocyte and osteoblast cells.

Such a challenge requires to assess multiple features of the cell dynamics and the reprogramming strategy from computational models of biological networks:
- impact of the kind of perturbations (persistent versus temporary) and of their order;
- inevitability of the targeted type after perturbation (fidelity);
- nature and duration of the triggered cascade of regulations (efficiency);
- robustness with respect to heterogeneity among cell population;
- robustness with respect to uncertainties in the computational model.
So far, no general framework allow to efficiently encompass those features to systematically predict best combinations of reprogramming determinants in distinct cellular reprogramming events.

In order to provide a scalable methodology, AlgoReCell will develop a theoretical framework based on the concurrency theory and abstract interpretation of discrete models of biological networks. Concurrency theory provides decompositions of systems dynamics by
exploiting the independence of the various sub-systems behaviours. It is particularly relevant for large gene regulation networks as the influences between those genes are typically sparse.
The assessment of fidelity and efficiency will be addressed by providing new probabilistic and stochastic models for a quantitative evaluation and ranking of potential reprogramming determinants.
Such a combination of computer science techniques has never been studied so far in the scope of cell reprogramming, and offers a very promising research direction.

The approach undertaken by the project will strongly support the emergence of an innovative theoretical-computer science assisted pipeline for cell reprogramming.
By building on top of formal approaches for dynamics of large networks, the project will bring a decisive step for the systematic prediction of reliable combinations of reprogramming determinants in distinct cellular reprogramming events.
Such a contribution will have a strong impact for the experimental cell reprogramming community, and in a broader extend, for regenerative medicine.

AlgoReCell relies on an original partnership between computational systems biology groups and theoretical computer science groups, with planned experimental validations. Such a collaboration aims at strengthening a fundamental base to the analysis of computational models driven by concrete, short or middle term, experimental challenges.

The application of AlgoReCell to the reprogramming of adipocytes and osteoblasts cells will allow to evaluate new strategies while collecting precious data for further understanding the undergoing differentiation process and refine computational models of this biological system.
We expect that these new predictions will improve the controllability of adipogenesis and osteoblastogenesis, which has important medical implications, notably for diabetes, obesity, and osteoporosis.