The international Global Alliance for Genomics and Health will be using a programming interface developed by the Internet giant to analyze genomic data.
Laura Landweber was 33 when she received tenure at Princeton. Oxytricha, beware: She's got a lot of science ahead of her.
November 1, 2006|
Her father is a mathematics professor at Rutgers. Her mom teaches English composition at Rider University. So perhaps it's not surprising that in 1994 Laura Landweber landed a faculty position at Princeton University - she was 26 years old. "I got an offer to start a lab and I was just too excited to defer that," she says.
Landweber hit the ground running. By the year 2000, she and her trainees had published dozens of papers on a diverse collection of systems related to biological computation, including RNA editing, the evolution of the genetic code, nucleic-acid computers, and the scrambled genomes of ciliates. When Stephen Freeland of the University of Maryland entered the lab as a postdoc in the late 1990s, he joined graduate student Rob Knight in studying the origin of the genetic code - a collaboration that Freeland says "stimulated some creative thinking" and was a credit to Landweber's "genuine interest in new ideas." In a series of 10 papers published in just two years, the researchers laid out evidence for why the structure of the genetic code was no evolutionary accident: The code, they argue, has been shaped by a chemical affinity between RNA codons and their associated amino acids, and optimized for its ability to buffer the cell against translational errors.
But Landweber might be most famous for her 2000 PNAS paper on biological computers, in which she used a test tube of RNA to solve a chessboard conundrum called the Knight's problem. In this poser, one attempts to find different ways to place a collection of knight pieces on a chessboard so that no knight can attack another. Taking advantage of RNA's penchant for hybridizing, and RNAse's penchant for digesting any RNA strands that don't find a partner, Landweber and her colleagues were able to isolate a set of "winning" molecules that describe the solution to the problem.
Her RNA computer was hailed at the time as the world champion by computer scientist Leonard Adleman of the University of Southern California. But Landweber subsequently chose to move on to other projects. "DNA computing is a difficult field to work in," she explains. "Silicon computers have had a 50-year head start, so the bar is set so high for a computation to be a meaningful one that it makes the field of wet computers pretty daunting. We didn't see the killer application, and most people in that field have shifted their interest toward building DNA molecules rather than using DNA molecules to solve problems faster than a computer can."
When she got tenure in 2001, Landweber decided to focus her attention on scrambled ciliate genomes. "Some people take the approach that before tenure they'll work on the one thing they do really well, and after tenure they might branch out," she says. "Somehow I did it in reverse." She credits, in part, her pedigree. Her mentors - Walter Gilbert and Richard Lewontin for her doctoral studies and Jack Szostak for her junior fellowship, all at Harvard - have diverse interests. "In Wally Gilbert's lab there were people working on zebra-fish development alongside me, working on trypanosome molecular genetics and evolution," she says. "So I was in an environment where people could work on anything that relates to work in the lab."
Her interest in ciliates (single-celled organisms found in oceans, soil, and pond water) was ignited when she heard David Prescott speak about his work with ciliates while she was still at Harvard. "I was immediately attracted to the problem," says Landweber, "and it was clear to me that gene scrambling was something I'd like to incorporate into the problems I'd attack later on."
The problem is this: Ciliates such as Oxytricha (Landweber's favorite) carry two separate genomes, one of which is a rearranged version of the other. The larger of the two contains about one billion base pairs (or one gigabyte of information), 95% of which is supposed junk. The second genome, about 50 megabases long, is a streamlined and decrypted version of the first genome and acts as the template for mRNA transcription. During development, the organism performs what Elaine Mardis of Washington University calls "a bizarre dance of genome trickery" to generate the descrambled genome.
To Landweber, the descrambling exercise is a good example of how nature solves a computational problem. "Executing the program that converts one genome, the larger germline genome, into the somatic genome that can be read by the cell and used to make proteins, well, that's computation." And computational biology is her real bread and butter. "One of the things I associate with Laura is her ability and enthusiasm and eagerness to look at biological problems and recast them as problems in mathematics or computation or logic," says Mark Ragan, head of the division of genomics and computational biology at the University of Queensland. "This is what we all aspire to do and it's something she does so well."
One thing Landweber learned by crunching the numbers is that the computational power of this particular form of genome trickery is equivalent to that of an electronic computer, says Lila Kari, a computer scientist at the University of Western Ontario who worked with Landweber to generate mathematical models of descrambling. If that finding seems surprising, it was even more so six or seven years ago. "Nowadays everybody takes for granted that biological organisms can do computation," says Kari. "Well, I have to tell you, when we talked with Laura in the beginning about this, people looked at us a little bit odd."
Landweber is also collaborating with Mardis and her colleagues on sequencing the descrambled genome of Oxytricha and, eventually, sampling the exon-rich regions of the one-gigabyte creature. Even the smaller genome is proving quite a challenge, as it comprises a collection of some 25,000 distinct chromosomes that range in size from single-gene 250 base-pair "nanochromosomes" to a slightly more conventional 35,000 base-pair eukaryotic chromosome. Although the project is daunting, Mardis says the collaboration has been great. "Clearly Laura and her lab care a lot about getting the genome sequenced and getting the sequence as good as it possibly can be, so that they and others in ciliate community can move forward using the data."
In the meantime, Landweber has done some limited sequencing in her lab in an attempt to study the molecular mechanisms that drive this biological encryption process and to explore how the system arose. One of the evolutionary studies involved a bit of fieldwork. "Folks in my lab started by just walking around Princeton and collecting a variety of different ciliate species from Lake Carnegie and the marshes in the surrounding area," says Landweber. "What we found was that, surprisingly, we could isolate a bunch of intermediate species that had differing levels of scrambled genes." The lineages that had diverged earliest had segmented genes, but the pieces were in the correct order. Later-diverging species started to show scrambling. "So we could basically trace a general trend toward an increase in scrambling complexity over the scale of evolutionary time."
"This gets at the deep questions of how organisms can do genetics and how genetics itself has been built up of component processes," notes Ragan. "After all, ciliates didn't invent all of the machinery to do these strange processes themselves." That's what makes the problem so much fun, says Landweber. "The odd and quirky genetic systems that nature has created over billions of years are not the solutions that one would design on paper if one were masterminding or engineering an elegant biological system. They're Rube Goldberg devices." Ultimately Landweber hopes to understand enough about how the systems function to be able to feed a ciliate an artificially scrambled construct and "observe the correct rearranged molecule as output from the cell. That will be our grail."
In the meantime, Landweber tries to attend meetings where, Freeland says, "she networks like crazy and spends time talking to people to find out what's going on and sees who's doing what, where." She encourages her students to do the same. "In the last four years I got to attend seven or eight national and international conferences," says Han Liang of the University of Chicago, a recent graduate. "This is unusual for graduate students. But Laura thinks that going to conferences allows students to meet people, exchange ideas, and really get to know the field."
She gives her students the same independence within the lab. "Laura really trusts and respects her students. She gives you freedom to do what you want as long as you can demonstrate your interest in a project and show that it's feasible," says Liang. "Laura thinks students should own their PhD experience."
When Freeland joined the lab, he more or less continued with the work he'd been doing as a graduate student. "Laura's strength is not so much imposing her projects on other people as it is noticing interesting work that's going on, bringing it in, and letting people meet one another and flourish," he says. "It almost put me in mind of a sort of Renaissance court where a patron would be fascinated with seeing science happen and would delight in bringing together groups of scientists and fostering their development. It's a rare skill and it's underappreciated."
There is one requirement, though: to sample all the best local chocolate. "Laura is like a professional chocolate taster. Whenever we go to a conference, she tells us which are the good chocolate stores in the city. So it's tradition for every lab member to bring some chocolate back when they travel."
And if they're anything like Land-weber, they'll also take time to dance. "I think I might have spread the dancing virus to Laura," laughs Kari, who shared her tales of ballroom dancing with Landweber as they hit the discos after the conference sessions. Landweber then took up swing dancing with her husband while on sabbatical at Caltech. They got good enough to earn a trophy, which "doesn't surprise" Mardis. "I don't think she does anything in a half-assed way."
Nowadays Landweber spends more time taking care of her two children than she does dancing and hunting down chocolate. In a way, that, too, was part of her plan. Landweber wrapped up her graduate work at Harvard in four years and did a one-year stint as a Harvard junior fellow with the hopes that such an "express" approach would allow her to better balance work with family. "It was my solution to how women can pursue a scientific career," she says. "Get through the academic hurdles as fast as you can to avoid the stress of having to start a family while you're trying to get tenured."
And though the hurdles have been cleared, the puzzles have not all been solved. "We are far from knowing everything that's happening," notes Kari, who continues to work with Landweber on refining the model of ciliate scrambling. "If Laura comes to us tomorrow and says, ?Whoops, I discovered there is a little genie in the ciliate. And the genie waves a magic wand and says, Unscramble,' okay, then we have to model the genie. The story never ends."
From Landweber's perspective, a never-ending story is a good thing. "One benefit of irreducible complexity and irreducibly complex problems in biology," she says, "is that they've raised a lot of really good challenges that people will be working on for the next 50 years."