Professor of Computer Science and Informatics
Going Small, Thinking Big
Randall Beer is using a tiny organism to help understand the human nervous system.
The quest to understand the human nervous system is a journey that has been pursued for centuries. But in order to discover the mysteries of the big stuff, sometimes you have to think small.
Randall Beer, a professor of computer science and informatics, believes understanding how a tiny worm with a long name functions on a neurological level can eventually help researchers better understand the human nervous system.
“We’re interested in trying to understand the behavior of Caenorhabditis elegans,” Beer says. “C. elegans is a one-millimeter long worm that is of great interest in biology. It was the first animal whose genome was sequenced, well before the human genome project was completed. In addition, the entire development of the worm is known. The adult worm consists of less than 1,000 cells and the sequence of cell divisions that take you from the fertilized egg to the adult worm have been worked out.”
Most importantly for Beer’s work, every neuron in the worm and the connections between these neurons is known. There are exactly 302 neurons and around 8,000 connections between them. No other animal’s nervous system has been mapped out to that level of detail. Even more impressive, new experimental techniques are allowing biologists to visualize neural activity in the worm’s entire nervous system as it behaves, as well as to manipulate the activity of individual neurons.
“Given this unprecedented level of experimental access and control,” Beer says, “C. elegans is an ideal target for computational modeling.”
Might it be possible to model an entire animal in a computer? He and postdoc Eduardo Izquierdo are attempting to do just that, one small step at a time.
“Our initial focus has been on modeling a kind of behavior called salt klinotaxis,” Beer says. “In this behavior, the worm crawls or swims up a gradient of salt concentration to its source, where the highest concentration of the bacteria on which it feeds can be found.”
They searched an online database of connections in the C. elegans nervous system for all pathways that connected chemically-sensitive sensory neurons to the motor neurons responsible for steering the worm’s motion. Then they simplified this initial network into a minimal circuit based on other experimental considerations and implemented it and a simple model of the worm’s body and chemical environment on a computer.
But there was one problem. Despite having a complete map of the worm’s nervous system, little is known about the properties of those connections.
“We may know that one neuron connects to another one, but we usually don’t even know whether it excites or inhibits it, let alone how strongly or weakly it does so,” says Beer.
In order to address this problem, they turned to evolutionary algorithms, a search algorithm based on biological evolution that allowed the computer to adjust the many parameters of their model.
It worked. The computer worm could find the bacteria as well as the real one. Furthermore, by running the evolutionary search multiple times, they discovered different parameter settings that were consistent with the biological data. By analyzing the operation of these alternatives, they were able to suggest specific biological experiments that might distinguish between them. In turn, the results of these experiments can be used to further constrain the next iteration of the model, resulting in a feedback loop between experiment and modeling that incrementally improves our understanding.
“Despite all the effort being put into the human brain, we don’t yet even understand the brain of a millimeter-long worm,” Beer says. “Working out the neural basis of behavior in a simpler animal like C. elegans might go a long way toward telling us how we should think about more complicated brains. Ultimately, ours.”