Cell Fate Forecasting: A Data-Assimilation Approach to Predict Epithelial-Mesenchymal Transition

Area of Science:

• Computational Biology
• Cell Biology
• Systems Biology

Background:

• Epithelial-mesenchymal transition (EMT) is crucial for development, regeneration, and cancer metastasis.
• Transforming growth factor-β (TGFβ) is a key inducer of EMT, driving cells from an epithelial to a mesenchymal state.
• Predicting cell state transitions computationally is challenging due to model parameter uncertainty and experimental noise.

Purpose of the Study:

• To demonstrate a data-assimilation approach for reconstructing cell states during TGFβ-induced EMT.
• To predict the timing of cell state transitions using limited, noisy experimental observations and computational models.
• To assess the feasibility and utility of data assimilation in forecasting cellular fate during EMT.

Main Methods:

• Utilized an ensemble Kalman filter for data assimilation, combining noisy observations with a computational model of TGFβ-induced EMT.
• Employed synthetic in silico experiments mimicking experimental parameter uncertainty and variability.
• Investigated effects of TGFβ dose, cell steady-state conditions, observation time intervals, and multiplicative inflation.

Main Results:

• Successfully reconstructed cell states and predicted future states in synthetic experiments, even with model error.
• Data assimilation proved effective when observations were frequent (short time intervals) and incorporated multiplicative inflation.
• The approach mitigated the influence of model uncertainty and error on predictions.

Conclusions:

• Data assimilation offers a feasible and valuable method for forecasting cell fate during EMT.
• This approach enhances the predictive power of computational models by integrating experimental data.
• Accurate prediction of EMT dynamics can be achieved despite inherent biological and experimental complexities.