Inferring parameters for a lattice-free model of cell migration and proliferation using experimental data. (21st January 2018)

• Record Type: Journal Article
• Authors: Browning, Alexander P.; McCue, Scott W.; Binny, Rachelle N.; Plank, Michael J.; Shah, Esha T.; Simpson, Matthew J.
• Is Part Of: Journal of theoretical biology. Volume 437(2018)
• Page Start: 251
• Page End: 260
• Publication Date: 2018-01-21
• Subjects: Individual based model -- Cell migration -- Model calibration -- Cell proliferation assay -- Approximate Bayesian computation; Biology -- Periodicals; Biological Science Disciplines -- Periodicals; Biology -- Periodicals; Biologie -- Périodiques; Theoretische biologie; Biology; Periodicals; 571.05
• DOI: 10.1016/j.jtbi.2017.10.032
• Languages: English
• ISSNs: 0022-5193

Abstract

Highlights: Lattice-free individual based model of a cell proliferation assay. Calibrate model using ABC with new experimental data for prostate cancer cells. Three independent experimental data sets produce similar parameter distributions. Calibrated model predicts an additional, fourth experimental data set. Experimental data highlights the role of clustering & spatial structure. Abstract: Collective cell spreading takes place in spatially continuous environments, yet it is often modelled using discrete lattice-based approaches. Here, we use data from a series of cell proliferation assays, with a prostate cancer cell line, to calibrate a spatially continuous individual based model (IBM) of collective cell migration and proliferation. The IBM explicitly accounts for crowding effects by modifying the rate of movement, direction of movement, and the rate of proliferation by accounting for pair-wise interactions. Taking a Bayesian approach we estimate the free parameters in the IBM using rejection sampling on three separate, independent experimental data sets. Since the posterior distributions for each experiment are similar, we perform simulations with parameters sampled from a new posterior distribution generated by combining the three data sets. To explore the predictive power of the calibrated IBM, we forecast the evolution of a fourth experimental data set. Overall, we show how to calibrate a lattice-free IBM to experimental data, and our work highlights the importance of interactions between individuals. Despite great care taken to distribute cells as uniformly as possible experimentally, we find evidence of significant spatial clustering over short distances, suggesting that standard mean-field models could be inappropriate.