Paper published in Cell Reports!
I am happy to share that our work on computational modeling of single neurons!
Mathematical models help us to understand how things work when we could not easily measure directly. For example, to simulate the galaxy structure we might never have the opportunity to measure stars that are millions light years away. Therefore many astrophysicists are using computer simulations to study what is happening really far away. In a similar fashion, we might not have the data to simulate all neurons in the human brain and measure all cells in every detail. There are 80 billions of cells and we would probably never measure the activity of every cell. Our work on making the computational workflow to generate single neuron models at scale aims to partially solve this problem. Using mathematical models it is possible to simulate new experiments and make better predictions into brain biology. We found that these models allow us to clarify the differences between different neuron types in the mouse cortex. There is also upcoming work on using this approach to study human epilepsy, but I will wait for it to be officially out.
In this work we found that computational models trained to electrical and structural properties of neurons can recapitulate the differences between cells. Now when we have the models of different neuron types, we could simulate realistic biological networks of cells and study theirs dynamics. One of the applications of these detailed neuron models are brain computer interfaces. Computer simulations of single cells and small networks also help us understand how electric fields are being generated in the brain. Brain computer interfaces that go inside the brain, like neuralink, use these electric signal to decode the neural activity and it is important to understand how it is organized biologically. Our generated models, as well as the raw data are freely available on the website for anyone who would like to use them for their own research.
Full version of the paper:
Official press release: