The human genome is a vast parts list for the inner works of our biology. It codes for thousands of proteins that make us who we are and keep our bodies up and running. Though that parts list was worked out a decade ago, its utility remains limited by the fact that we still don’t have a wiring diagram to go with it. That is, we don’t know how all the parts interact.
“This is where true richness of phenomena originates,” said Baruch Barzel, a postdoctoral researcher in the Center for Complex Network Research, the lab run by world-renowned network scientist and Distinguished University Professor Albert-László Barabasi. “Disease mechanisms, as well as our healthy functionality, are all encoded not just in the genes and proteins, but also in how they interact to form a network.”
But unlike many networks—like the components inside your car engine or the wires inside a mechanical doll—biological systems are black boxes. We can observe the outcome of their interactions, but not the interactions themselves.
In work published online Sunday in the journal Nature Biotechnology, Barzel developed a mathematical method for peering inside that box. The research moves the team a step closer in its quest to understand, predict, and control human disease.
It is estimated that a mere 0.01 percent of all possible interactions in the human genome actually take place, Barzel said. But with 20,000 genes, that still leaves room for up to a 100,000 interactions. Since a blind search for these would be infeasible, scientists have developed high-throughput experimental methods to efficiently detect them. The only problem is, once again, that black box.
The experimental data provides the global picture of the cell after all interactions have taken place. For instance, if protein A affects protein B, which in turn affects protein C, the link between proteins A and C would also register. Barzel’s method allows him to “silence” those indirect links.
“After silencing, what you are left with is the precise wiring diagram of the system,” said Barzel. “In a sense we get a peek into the black box.”
To test the method, Barzel applied it to a network of 4,511 Escherichia coli genes. He found that the approach performed up to 67 percent better at identifying real interactions than standard correlation methods.
But their wiring diagram is just the beginning. “You don’t only want to know which component acts on which other component, but also how they act on each other. What is the mechanism by which their interaction is realized,” said Barzel. “This is a completely fresh problem in the area, and our method can help solve it.”
The research continues Northeastern’s leading work in network science, as its researchers are tackling a range of projects in this emerging field that involves understanding the complexity that governs all systems. This work spans from studying the global spread of disease to analyzing social media data as a way to better understand fields ranging from political science to disaster preparedness.
Northeastern also announced the launch of the nation’s first doctoral program in network science. The university will begin recruitment in the fall.