Reverse-engineering brain neurons to build better computers

Sep 18 2007

Researchers at the University of Texas at San Antonio (UTSA) are using Star-P software to reverse-engineer brain neurons in a quest to build better computers.

A team within UTSA's biology department is taking advantage of powerful parallel computers to run biologically-realistic simulations of molecular diffusion in neurons. By understanding how neurons process chemical signals when a person learns and remembers information, researchers believe they can create more reliable computers that employ stochastic computing components. Stochastic computing is a type of artificial intelligence which uses probabilistic methods (i.e. chance) to solve problems.

UTSA's work could also lead to other neurobiological research breakthroughs, particularly in realms of sensory acquisition, motor learning, disease, and higher cognitive functions.

Led by Fidel Santamaria, Ph.D., assistant professor of computation and neural systems, the UTSA team created a computational and experimental lab that integrates electrophysiological, imaging, and structural observations of neurons into detailed biophysical models. Since the human brain has trillions of different types of neurons, each with complicated branching dendrites, running the complex Monte Carlo simulations to model even a single neuron requires enormous computational performance and memory resources.

To this end, the department purchased a Star-P(R) license to link their desktop computers to an 8-processor parallel cluster, with funding support from the Cajal Neuroscience Research Center and the National Institute of Health. To further accelerate the team's research, Interactive Supercomputing granted UTSA an additional license to deploy Star-P on a 120-processor cluster in the near future.

Born and used extensively in MIT's labs, Star-P is an interactive parallel computing platform that enables the team to code the biophysical models on their desktops using MATLAB(R), but run them instantly and interactively on the parallel cluster with little to no modification. Star-P eliminates the need to re-program the applications in C, Fortran or MPI to run on parallel systems, which would take months to complete for large, complex problems such as UTSA's molecular diffusion simulations. The result has been dramatic improvement in research productivity.

Santamaria said that the team has doubled its productivity by breaking up the models into smaller data sets and running them on the supercomputers with Star-P. "They can now do eight analyses at the same time," he said. Even for larger simulations of entire neurons, Santamaria said he has gained a 50 percent increase in productivity using Star-P. "Not only that, we can model much more complex molecular structures compared to what we could do before. While simulating a single neuron may not sound like much, Monte Carlo simulations of molecular diffusion in spines and dendrites is an enormous computational challenge."

In additional to the team's stochastic computing research, Santamaria said he plans to utilize Star-P as a student lab tool for a computational neuroscience class that he teaches. The department also intends to integrate Star-P into its Computational Biology Initiative (CBI), a new interdisciplinary initiative at the University of Texas Health Science Center at San Antonio (UTHSCSA) and UTSA. The interdisciplinary team using Star-P will include members from the University's civil engineering, computer science, math and neurobiology departments. The CBI's goal is to build infrastructure to significantly advance collaborative interdisciplinary bioscience research in San Antonio.

"Using state-of-the-art computational tools for uncovering the secrets of the brain, stochastic computing research at UTSA will contribute to not only better computing methods, but to our understanding of learning and memory, as well as neuropathologies such as Alzheimer's Disease, Parkinson's Disease, schizophrenia, and epilepsy," said Ilya Mirman, ISC's vice president of marketing.