Research

Biological systems are governed by emergent phenomena from interactions. I am interested in learning from real data those models of interactions. I enjoyed working on the following problems: How do species interact after mediation from the food source [1]? How do neurons interact and drive locomotion in zebrafish [3] and C. elegans [5]? How do mice interact and give rise to social behaviors [2]? With data-driven interaction models, we can make further predictions on controllability of those networks [4]. 

[1] X. Chen, K. Crocker, S. Kuehn, A. M. Walczak, T. Mora, "Inferring resource competition in microbial communities from time series," arXiv:2501.04520 (2025). [link]

[2] X. Chen, M. Winiarski, A. Puscian, E. Knapska, T. Mora, A. M. Walczak, "Modeling collective behavior in groups of mice housed under semi-naturalistic conditions," bioRxiv (2024). [link]

[3] X. Chen, F. Ginoux, M. Carbo-Tano, T. Mora, A. M. Walczak, C. Wyart, "Granger causality analysis for calcium transient in neuronal populations: challenges and improvements," eLife (2023). [link]

[4] E. D. Lee, X. Chen, B. C. Daniels, "Discovering sparse control strategies in neural activities of C. elegans," PLOS Comput Biol (2022). [link]

[5] X. Chen, F. Randi, A. Leifer, W. Bialek, "Searching for collective behavior in a small brain," PRE (2019). [link]

A key step of understanding emergent collective behavior in living systems is to build realistic models for interactions, which has unique challenges as living systems are heterogeneous, out-of-equilibrium, and having many timescales. In order to address those challenges, I develop novel methods of statistical inference, such as the generalized Glauber dynamics that keeps the steady state the same while explore a large class of dynamical models, and applied it to study the collective dynamics of social mice [6].

[6] X. Chen, M. Winiarski, A. Puscian, E. Knapska, A. M. Walczak, T. Mora, "Generalized Glauber dynamics for inference in biology," PRX (2023). [link]

A hypothesis for a theory of living system is that they are tuned close to criticality, such that they maximize the performance of their biological functions, whatever it means. I ask questions such as how easy it is for a network to achieve such criticality, a.k.a. whether there is fine-tuning [7]. A while back, I also looked at the connection between the geometry and dynamics in systems that exhibit transient chaos [8].

[7] X. Chen, W. Bialek, "Searching for long time scales without fine tuning", PRE (2024). [link]

[8] X. Chen, T. Nishikawa, A. E. Motter, "Slim fractals: the geometry of doubly transient chaos, " PRX (2017). [link]