Neuroscientists and computer giant IBM collaborate to build the first simulated model of human brain circuitry in what is known as The Blue Brain Project.

How the human brain works has always been a great mystery, eluding and baffling neuroscientists through the ages. At the center of this mystery has been the long quest to determine the structure and function of the most complex area of the human brain—the neocortical column, or neocortex. An area of the brain that allows mammals to adapt rapidly to a changing environment, the neocortex was initially investigated by the now-famous Spanish neuroscientist, Santiago Ramon y Cajal.

“Back in the 1800s, Cajal described the neocortex as a jungle, and he advised students and researchers to stay away from it,” says Charles Peck, PhD, manager of

This figure illustrates the synaptic complexity of a single neocortical minicolumn, the framework of the human neocortex. (Source: EPFL)

biometaphorical computing research at IBM, Seattle. Regardless of this complexity, Cajal fervently continued his mission to reveal the microstructure of the brain. And, in 1906, he received the Nobel Prize in Medicine for his groundbreaking work, which formed the basis of today’s neocortex research.

Much work has been accomplished since 1906, however. “Over the years, a lot of data have been accumulated on the neocortex, in terms of what kinds of cells it possesses, their distribution, how they are connected to each other, and so on,” says Peck. The heterogeneity of this particular region of the brain has also been observed, and scores of research reports document the vast morphological and electrophysiological differences between resident neurons.

The neocortical column, a collection of neurons connected to each other in a very complex way, represents the smallest network of neurons to be considered a functional unit. The column is composed of six layers, with each layer sending and receiving nerve impulses to and from different regions of the brain. The neocortex is also composed of hundreds of mini-columns, each with its own function. Because of this complexity, the column marks the evolutionary leap from reptiles to mammals, and, therefore, may explain human intelligence and cognition.

Although a lot of progress has been made to understand these features, the overall function of the cortex is not well understood. Enter The Blue Brain Project. According to Peck, the biological purpose of the project is to build a computer model of the underlying structure of the neocortical column and simulate the physiology, such that a better understanding of brain function and cognition can be attained.

A new path
The Blue Brain Project was officially launched in June 2006. However, the inception of the project occurred at least 10 years ago when Henry Markram, PhD, a professor at the Ecole Polytechnique Fédérale de Lausanne (EPFL) and mastermind behind the project, started to piece together the neuronal circuitry of the neocortical column. Markram could not be reached for an interview for this story.

Markram’s former graduate student, Yun Wang, MD, PhD—now a research assistant professor at Tufts University School of Medicine—did much of the work that formed the basis for the project as part of his doctoral thesis. “For my PhD work, I did a lot of reconstructions of the neuron connections. This basically became the database for The Blue Brain Project,” says Wang.

With these data in hand, Markram went out to solicit funding for the project, but the problems he had in getting the project started were not all financial. He talked to many people at IBM to work out a potential collaboration with the EPFL. “Henry Markram had been interested in this project for a while, and the project had a sufficient amount of data to get started, but the problem was that until The Blue Gene supercomputer became available, there really wasn’t any kind of computational data available to even approach the problem,” says Peck, who heads this project for IBM. Markram was later able to convince IBM to support The Blue Brain Project, obtaining tens of millions of dollars for it.

According to the project’s Web site, the mission is to “understand mammalian brain function and dysfunction through detailed simulations.” Of course, this is the overall objective of the project and very long-term. The first thing the researchers will do is use The Blue Brain supercomputer to build an accurate, software-generated replica of the neocortical column.

“Henry has big plans for this project,” says Wang. “He is building a cortical column of the somatosensory cortex. Once this column is built, it can be modified for other columns.” For example, if Blue Brain researchers can determine the organizational principles of the cortical column in the prefrontal cortex, then they will use the knowledge obtained from this single column model to predict the function of the entire cortex.

Before the availability of the supercomputer, Markram was performing 12 simultaneous patch clamp experiments on a single slice of neocortex and, although that technique is state-of-the-art, the information provided by those experiments was a little sparse, says Peck. “However, simulations allow us to model every single neuron, say 10,000 neurons in a neocortical column at a very high level of detail, both in terms of structure and electrophysiology. So, with simulations we have access to the entire state of the system.” Markram anticipates that The Blue Gene supercomputer will significantly accelerate the time for running real-time neuroscience experiments to the point where a full day’s worth of wet lab experimentation may take only seconds because much of the pre-planning can be done in silico.

A majority of the pre-planning steps involve adjusting the conditions or parameters of the experiment. There are hundreds of thousands of parameters in the simulation model that cannot be observed biologically. Consequently, these researchers have to develop tools and techniques to get to those parameters. “And so what that involves is starting off with models of parts of the system, validating the parameters of those experimentally, and then building up from those,” says Peck.

A typical Blue Brain experiment starts off with a single neuron. It is verified that the neuron’s function can be replicated in silico. This is repeated with several types of neurons. Following this simulation, all of the neurons are joined into small networks, and the networks validated to determine whether the results match with in vivo activity. And according to Wang, Blue Brain researchers also try to replicate neuronal behavior from the top down. For example, they take measurements from tissue sections of neocortex from an animal, observe the emerging behavior, and then try to replicate that behavior with The Blue Gene supercomputer.

Different shades of blue
There are also plans to look at the neocortical column at the genomic and proteomic levels. This is possible because of the simultaneous patch clamping technique. After clamping onto a single neuron and recording its electrophysiological properties, the same probe can extract DNA to do genomics. Finally, a dye is injected into the cell, and images of the cell’s morphological features can be taken.

“Henry Markram’s group looked at very specific genes related to ion-channel control and expression, and there was a very strong relationship between the electrophysiological behavior of a neuron and expression of these specific genes,” says Peck. “This means that it becomes possible to predict the electrophysiology of individual neurons based on their gene expression profiles.” In 2007, The Blue Brain Project will begin to do just that.

Furthermore, because the parameters of the electrophysiology of neurons in a given circuit can be manipulated using The Blue Gene supercomputer, it is possible to perform a perturbation analysis on that circuit. For example, if a change in a parameter yields a large effect on the electrophysiology of that circuit and the emerging phenomena, then altering that parameter pharmacologically might lead to beneficial effects, such as treating a disease. Thus, simulation of neuronal function of the neocortex will certainly aid in designing drugs to treat neuropsychological diseases, which is, according to Peck, the ultimate goal of The Blue Brain Project.

This article was published in G & P magazine: Vol. 6, No. 9, December, 2006, pp. G10-G11.