Wednesday, November 22, 2006

Brain wiring patterns

Gnome writes "Complex electronic circuits designed by humans are composed of many repeated, simpler modules such as amplifiers or logic gates. Does the brain have a similar organization? Reigl and colleagues looked for such patterns in C. elegans, a worm whose nervous system wiring of its 302 neurons could be analyzed from electron micrographs of serial sections. Here, they analyzed the frequency of connection patterns up to 5 neurons. [They concede that their analysis would miss singular circuits, such as a crucial “rectifier in a power supply” (returning to the electronic analogy).] Nevertheless, they found that certain connection patterns between 2, 3, and 4 but not 5 neurons occurred more often than chance. For example, a triplet where 1 neuron stimulates 2 neurons, one of which stimulates the other, occurs much more often than expected. They conclude that these motifs could perform stereotypical functions in the worm nervous system. Many useful circuits need many more connections.
Full&Free Reigl et al., BMC Biology 2004, 2:25
Search for computational modules in the C. elegans brain


Reuel said...

Synthetic Biology

Synthesizing circuits should be nicely complementary to analyzing circuits. Wired has an article [] on a class at MIT that is building small DNA-protein circuits that do things like count (e.g., number of divisions) or blink (on-off).

Anonymous said...

Wrong Random

Their null hypothesis network was created by randomly connecting cells. This produces a normally distributed population of connections with most cells having near the average number of connections. However, biological networks (and other naturally growing networks such as the internet) display a power-law distribution of connections [] with few, very well connected nodes. A familiar example of power-law distribution is the hub-and-spoke pattern of airline connections whereas a more normal distribution is the interstate highway pattern connecting cities.

Their null hypothesis should be tested against this more natural, but still random network. In a Science Note [] on a similar paper by Milo et al. [],
Artzy-Randrup and colleagues claim that random networks so constructed have many characteristic patters, such as feed-forward circuits, that are not common to random networks with normally distributed connections.

It should be possible to filter the circuits found in the C. elegans central nervous system
to eliminate those resulting from the power-law alone.