Edward Marcotte is using networks—like those that could be used to illustrate the connections among users of the Web sites MySpace and Facebook—to study life itself.
Edward Marcotte is using networks—like those that could be used to illustrate the connections among users of the Web sites MySpace and Facebook—to study life itself, creating maps of the hundreds of thousands of millions of potential relationships among genes and proteins within an organism.
“We use networks because they are a very general way to visualize complex relationships,” says Marcotte, professor of chemistry and biochemistry. “How would you picture all of the structure in Facebook.com? All of the cliques, friends and family? A network is a very powerful and natural way to think about those relationships.”
Marcotte has already used networks to show how some genes are related through evolution. He’s working on a project now that may give scientists an idea of which of the 20,000 or so human genes are essential for life. His networks also offer a brand new way to predict the effect of a new drug, find the root cause of disease, and discover new genes.
The networks could, in fact, revolutionize genetics. Scientists may have sequenced the human genome back in 2003, but we still know very little about what all of the genes do.
“Even in yeast, one of the best studied organisms on the planet, we still don’t know what roughly one third of the genes are doing,” explains Marcotte. “In humans, we know what less than half the genes are doing. We know they are there, we know the sequence, but we don’t know how they act in the organism.”
Oh, and by the way, his networks have crept into the art world, too. A rendering of Marcotte’s “Protein Homology” network appeared this spring at the Museum of Modern Art in New York as part of their “Design and the Elastic Mind” exhibit.
A visualization of a gene network looks like a satellite view of Earth’s cities clustered and sparkling across the night-darkened landscape. Genes are the little bright nodes, and if they have a high probability of working together in the cell, they are connected to each other by a line, or “edge.”
Similar genes cluster together into neighborhoods, and these neighborhoods form denser and denser clusters the more related the genes are to each other or the more similar the genes’ function. Zoom out and you can see the general patterns of gene relationships. Zoom in and you can explore each of the genes within a neighborhood by following the roads that connect them.
“You can think of it like six degrees of separation or a Facebook.com for genes,” Marcotte says. “If you know of a few genes and what they do, their ‘friends’ probably do something similar, and we can find these through the network.”
A self-avowed computer hacker since he was a kid, Marcotte and his team began developing software to connect and display genes in networks about a decade ago. The open-source software has since been used by others to, among many things, map the connections of the Internet and cell phone networks.
Marcotte recently received a cryptic email from the U.S. National Security Agency thanking him and his colleagues for developing such nice software. “We don’t know how they’re using it, and we’re not asking,” Marcotte says with a grin.
Thus far, Marcotte and his team of computational and experimental biologists have used their software to build networks for the workhorses of genetics: yeast, the mouse and the nematode worm, Caenorhabditis elegans. Two other networks—one for the mustard plant, Arabidopsis, and one for the human—are in development now. “Those two are not quite ready for prime time,” he says, but the first three are fully available and online at functionalnet.org.
This past year, Marcotte and his team reached a major milestone by proving that their nematode worm network can be used to find new genes and predict their function. In a January 2008 paper published in Nature Genetics, Marcotte and then post-doc Insuk Lee showed how they could identify genes that regulate tumor development in the nematode.
Lee synthesized data from about 20 million experiments from around the world to build a network, called “Wormnet,” connecting the worm’s 16,000 genes. He and Marcotte explored the Wormnet and identified about 170 genes that had a high probability of being involved in the development of tumors. The tumors are a model for human eye cancer (retinoblastoma) and appear as growths along the length of the worms’ bodies.
To test their predictions, they collaborated with Andrew Fraser’s group at The Wellcome Trust Sanger Institute in Cambridge, U.K. There, researchers inactivated the candidate genes with a technique known as RNAi, which basically mimics the action of a drug by knocking out the function of individual genes.
Fraser says that RNAi allows them to remove any single gene and see what effect it has on the entire organism—an advantage the worm has over studying single human or mouse cells. His group found that inactivating 16 of the 170 genes reversed the tumors.
“This showed us that we can use the network to accurately predict the function of genes,” Fraser says.
A major advantage of using networks, he says, is that they pull enormous amounts of genetic data generated around the world into one place, and they give researchers an efficient way to find new genes involved in any process. “If we only know of a small handful of genes involved in a process, we can more efficiently predict other genes involved,” he says.
As another proof-of-concept, the Marcotte team used the Wormnet to identify new genes that could play a role in muscular dystrophy. Ben Lehner, a post-doc in Fraser’s lab, then tested the new genes in the worms and discovered that they indeed interacted with known muscular dystrophy genes.
“The genes strongly altered some very basic, very important cellular pathways involved in the developing worm,” says Marcotte. “Those pathways should be investigated in muscular dystrophy patients to see if they are also involved in humans.”
The worm might seem a far cry from a human, but both Marcotte and Fraser say that discovering genes using the human network will work in the same efficient, friend-of-a-friend way as it does in the worm.
“If you know a few genes that cause a disease with a genetic basis, like Type II diabetes,” says Marcotte, “you can look at their neighbors in the network, and those genes become likely additional candidates involved in the disease.”
Marcotte and John Wallingford, a University of Texas at Austin developmental biologist, are beginning to look for new candidate genes involved in a number of human diseases, including spina bifida.
The second most common birth defect, spina bifida occurs when the neural tube doesn’t seal properly in the developing embryo. Babies are born with an exposed spine or brain case. The debilitating disease, like many diseases, is probably caused by the complex actions of many hundreds of genes.
The two professors’ labs work in sequence. First, the Marcotte lab uses the mouse and human gene networks to identify candidate genes for neural tube disorders. Then the Wallingford lab tests the function of equivalent genes in the Xenopus frog, which is a good model organism for studying neural tube development because the embryos of frogs and humans—and most vertebrates—share the same basic instructions for making tissues.
The frog embryos develop fast, and within a day, Wallingford and his group can see if the genes in question play a role in the sealing—or lack thereof—of the neural tube.
“We can test this very rapidly in the frog system,” says Marcotte. “Any genes that we find that we can implicate in this process, we can then look for among human spina bifida cases and start thinking about developing diagnostics for the disease.”
Wallingford and Marcotte are also looking for the genes that turn on and off in developing embryos to make cellular cilia, the microscopic hairs that line the moist insides of our lungs. Malfunction in the cilia is associated with asthma.
And in another collaboration, Marcotte is using networks to explore genes involved in cancer metastasis with University of Texas at Austin biomedical engineer Muhammad Zaman.
In a bigger picture way, the young and very busy Marcotte (always a coffee in hand it seems) wants to find the genes that are essential for a human cell to live.
“The network structure itself can tell us whether a gene is essential,” he says. “Nodes at the center of the network tend to be much more critical for the function of the organisms. Genes at the periphery tend to be outsiders—less essential. Essential genes are friends of other essential genes; they cluster together.”
Genes play center stage in Marcotte’s work, but he could just as easily be talking about the social life of college students, the patterns of the blogosphere, or even life on the streets in a crime show like HBO’s “The Wire.” It is through exploring the tangled links within such complex networks that minor players lead to major players and the kingpins are flushed out.
The big discoveries, and the stories that lead to them, emerge from the network.
This article also appeared in the Spring 2008 issue of Focus magazine.
Comments