Bio papers — 2026-10-21
The Morning Preprint — this day as one edition: the day's papers, the five things worth knowing, and the episodes behind them.
Today's focus is on how we are trying to better understand calcium signaling dynamics in cell culture using data-driven modeling because getting a clearer picture of these processes allows us to observe and characterize biological systems with greater detail. We are comparing two methods for system identification and parameter discovery for governing ordinary differential equations, specifically the Constrained Regularized Least-Squares Method (CRLSM) and Physics-Informed Neural Networks (PINN) in the work titled Data-driven multi-agent modelling of calcium interactions in cell culture: PINN vs Regularized Least-squares.
The core idea here is to use data to discover dynamics, which is powerful because classic methods often require us to already know a good set of candidate terms, which isn't practical for real biological studies. The CRLSM achieved a fairly good parameter estimate and a good data fit when using the learned parameters in the Consensus problem.
However, despite the initial hypothesis that PINNs would perform well, they failed to match the CRLSM performance and did not provide fair parameter estimation under their current setup. This suggests that while PINNs are promising, there is still significant room for improvement through more careful tuning and uncertainty quantification in future research.
Today's papers
- Data-driven multi-agent modelling of calcium interactions in cell culture: PINN vs Regularized Least-squares Data-driven discovery of dynamics in biological systems allows for better observation and characterization of processes, such as calcium signaling in cell culture. This paper compares two methods, PINNs and regularized least squares, to find the best way to model calcium delivery in cells. [paper] [episode]
The papers
- Data-driven multi-agent modelling of calcium interactions in cell culture: PINN vs Regularized Least-squares — The gist The CRLSM achieves a fairly good parameter estimate and a good data fit when using the learned parameters in the Consensus problem Introduction and Motivation Fundamental cells mechanisms such as excitation-contraction and gene expressions result from the Ca2+ intracellu [episode]
Important terms
- Calcium signaling dynamics
- This refers to how calcium ions move and change within cells during biological processes. Understanding these changes is key to seeing how biological systems work in detail.
- Data-driven modeling
- This approach uses real experimental data to build mathematical models of cell behavior. It's powerful because it lets us discover unknown dynamics instead of guessing the equations first.
- System identification
- This is the process of figuring out the underlying mathematical rules or parameters that govern a system based on observed data. It helps us characterize biological processes.
- Constrained Regularized Least-Squares Method (CRLSM)
- A specific mathematical technique used to estimate parameters in differential equations. It performed well here by giving good parameter estimates and a good fit to the experimental data.
- Physics-Informed Neural Networks (PINN)
- A type of neural network that incorporates the known physical laws (like differential equations) directly into its training process. They are promising but need better tuning for this work.