Dec 04, 2023
FitMultiCell – Parameter estimator for data-driven modelling
In fundamental biological research and in biomedical applications, multiscale simulations of cell-cell interactions in tissues are used, which are inherently stochastic and therefore cannot be classically optimised. Until now, it has been very unstable and computationally expensive to determine the required, but not directly measurable, model parameters by comparing simulations with microscopy images and videos. With the open source software "FitMultiCell", a user-friendly and robust parameter estimator for data-driven modelling has now been developed in a collaboration between the ZIH and the University of Bonn and published in the renowned journal Bioinformatics.1 (https://doi.org/10.1093/bioinformatics/btad674).
The FitMultiCell software has already proven itself in numerous practical applications. For example, Prof. David and his team of developmental biologists in Paris are delighted: "By combining high-resolution microscopy with FitMultiCell, we were able to discover the collective mechanism 'guidance by followers' during zebrafish embryogenesis." (Boutillon et al., Developmental Cell, https://doi.org/10.1016/j.devcel.2022.05.001)
Illustration of 2 interlocking workflows in FitMultiCell. On the left the iterative optimisation process is shown. The right one focusses on one of the iterations: combinations of test parameter values are drawn from the probability distribution. They are inserted into the computer model and simulated. Simulation results and measured values are compared. Depending on the result, the probability distribution is updated and the iterations are continued.
At its core, the open source modelling software "Morpheus", which is widely used worldwide and was developed at the ZIH department for Innovative Methods of Computing (IMC), was coupled with Bayesian parameter inference and model selection algorithms and the new data standard "PEtab-MS" was established for this large problem class (https://gitlab.com/fitmulticell/fit). Also the documentation is very user-friendly, with many application examples. The BMBF-funded project FitMultiCell (funding code: 031L0195B), under the coordination of Prof. Hasenauer at the University of Bonn, has been successfully completed with these results.
Contact
Fußnoten
-
E. Alamoudi, Y. Schälte, R. Müller, J. Starruß, N. Bundgaard, F. Graw, L. Brusch, J. Hasenauer. FitMultiCell: Simulating and parameterizing computational models of multi-scale and multi-cellular processes. Bioinformatics 39 (11), btad674, 2023. https://doi.org/10.1093/bioinformatics/btad674