Dear Brian2 developers and users:
I am glad to announce that my newest study using LAMMPS has been published on Journal of The Royal Society Interface: https://royalsocietypublishing.org/rsif/article/23/239/20251155/482038.
This work established an integrated mechano-physiological framework (CREST) for closed-loop control of the small intestinal transport. I used Brian2 to simulate the control function of the enteric nervous system. A biophysical neural network was constructed with physiological details, including the ion channels and synapses. In the CREST framework, Brian2 received mechanical stimulation from the physical simulator (LAMMPS), activated sensory neurons, interneurons, and then motor neurons. The motor neurons sent spike trains to the smooth muscle cells, and the latter generated active force to drive the intestinal musculature to move.
Here you can get access to my source code: Jasmine969/CREST_SUDA
I want to thank @mstimberg for his kind help. He replied to almost every question posted by me and always provided useful advice, saving me a lot of time.
There are currently many neuron simulators, e.g., NEURON, Brian2 and BrianPy. I tried all of them in the early stage and finally chose Brian2. In my work, I needed to run the Brian2 simulation for a certain period of time, pause it, wait for the LAMMPS calculation, obtain the updated mechanical stimulation, and then continue the Brian2 simulation. This made me appreciate Brian2’s Cython mode very much. After the first compilation, the generated code was cached, so restarting the simulation became much faster in the subsequent coupling steps. In comparison, BrainPy relies on JIT compilation, and in my coupled simulation workflow each restart introduced a large overhead. NEURON is also a very powerful simulator, but it would have required me to learn a new modeling language, NMODL, which has a relatively steep learning curve for beginners. Therefore, among these three simulators, I found Brian2 to be the most suitable and user-friendly choice for my project.
More importantly, Brian2 allowed me to focus on the scientific problem rather than spending too much time on technical details. The model equations could be written in a clear and readable form, and the simulation workflow was flexible enough to be coupled with an external physical simulator. This flexibility was essential for building the CREST framework, where neural control and mechanical transport had to interact repeatedly in a closed loop.
I would also like to share a broader thought from this work. In neuroscience, much attention has naturally been devoted to the brain and the central nervous system. However, the enteric nervous system, often referred to as the “second brain”, is also a highly autonomous and functionally rich neural system. It can sense the mechanical and chemical state of the gut, process information locally, and generate coordinated motor patterns such as peristalsis with limited central supervision. Despite this complexity and importance, the enteric nervous system still receives much less attention than the brain.
I hope that this study provides a useful example of how computational neuroscience tools can be applied to the gut. By coupling a biophysical enteric neural network with a physical simulator of intestinal transport, we can study not only neural activity itself, but also how neural control gives rise to organ-level physiological function. I believe this type of closed-loop, system-level modeling may help bridge neuroscience, physiology, biomechanics, and engineering, and may encourage more researchers to explore the nervous system beyond the brain.
Thank you again to the Brian2 community for developing and maintaining such a useful tool. I hope this work can serve as one example of how Brian2 can be used not only for neuroscience simulations, but also for interdisciplinary studies combining neural control, biomechanics, and soft-matter transport.
Here is a video showing the coupled dynamics of the intestinal transport and neural control: Item - Intestinal transport under neural control from Closed-loop control of the small intestinal transport within an integrated mechano-physiological framework - The Royal Society - Figshare
Cheers,
Hong Zhu