 |
Computational artist extraordinaire:
A conversation with BioE’s Kimmen Sjölander
by Gordy Slack
photos by Peg Skorpinski
 |
Bioengineering
professor Kimmen Sjölander joined the Berkeley Engineering
faculty in 2001. She develops computational methods that reveal
evolutionary relationships among proteins, the workhorses
of all life.
PEG SKORPINSKI PHOTO |
On the wall above Professor Kimmen Sjölander’s desk
hangs a photograph of the Beatles, taken at the release party
for their Sgt. Pepper’s Lonely Heart’s Club Band album.
They are beaming, soaring at the top of their game. Sjölander
keeps that photo to remind herself and her students that creative
play and hard work make beautiful music together.
Sjölander made her mark early. Even as an undergraduate at
UC Santa Cruz, she was developing innovative algorithms under
the wing of bioinformaticist David Haussler. Today, a decade later,
her digital code is all over the field of computational biology,
particularly in the area of protein phylogenomics. In September
she won a Presidential Early Career Award for Scientists and Engineers
(PECASE), the nation's highest honor for scientists in the early
stage of their careers.
In 1997, Sjölander went to work for a biotech startup that
was soon bought by Celera Genomics, where she played a key role
in the functional analysis of the proteins encoded in the human
genome. As principal scientist for the Protein Informatics group
at Celera, she coauthored the landmark publication of the human
genome in the journal Science.
For the past three years, Sjölander has been on the College’s
bioengineering faculty heading up the Phylogenomics Group. Her
lab develops computational methods that uncover evolutionary relationships
among proteins, allowing scientists to infer the structure and
function of newly discovered proteins on the basis of their relationships
to known ones. Identifying the structures and functions of proteins—the
workhorses of all life— helps biologists untangle the story
of evolution and will be key to helping genetic researchers understand
how proteins confer disease resistance in plants and animals,
and perhaps in the development of new medicines.
A faculty scientist at Lawrence Berkeley National Laboratory,
Sjölander is a member of the three-campus initiative, the
California Institute for Quantitative Biomedical Research (QB3).
She holds a joint appointment in the Department of Plant and Microbial
Biology, where she works with biologists on plant disease resistance.
Using a National Human Genome Research Institute grant, she is
also developing a digital catalog of human brain proteins.
Q: Let’s dive right in and
talk about how you launched your career in computational biology.
K.S: My bachelor’s degree and my Ph.D.
at Santa Cruz were both in computer science. I focused on machine
learning, which comes down to developing basic mathematical and
algorithmic approaches to extracting information from data. I
had the great luck to work with [UC Santa Cruz computer science
professor] David Haussler, who was starting to apply machine learning
to biological data. Based on a research project I did as a freshman,
David invited me to be part of a team applying methods from speech
recognition to the problem of multiple-sequence alignment of proteins.
This ended up being the genesis of hidden Markov models for proteins,
which had an enormous impact on computational biology. It took
off from there.
Q: It sounds
like you walked right onto a rocket ship that was already smoking
on the launch pad.
K.S: Totally. I had dreamed of doing research
and being a scientist, and suddenly there I was. Those were amazing
days, full of creativity. First thing I did in the morning was
turn on the computer to see what happened to my experiments the
night before, and I haven’t stopped. I’m as excited
and obsessive as I was back then. On the other hand, it wasn’t
easy. When I went back to school, I had three young kids. My twins
were two and my oldest was five.
Q. What was it like for your children to grow up when
you were so heavily into your studies?
K.S: I chose to do my undergraduate work at Santa Cruz
partly because it provided such a healthy framework for raising
my children. Family student housing at Santa Cruz was very supportive.
Being a single parent is really hard for most people, and it wasn’t
easy for me, but for the most part we had a tremendous time. There
might have been times when my kids wished they had a mom who’d
have cookies and milk on the table when they got home, but not
many. We did a lot of hiking and exploring and gathering lichen
in the redwoods. We still think about how great those times were.
Q: You went back to school after a 15-year academic hiatus.
You explored a lot of different worlds during that period. Tell
us about your life between leaving school and returning to Santa
Cruz as an undergraduate in 1990?
K.S: My twin sister and I started college at City College
of New York when we were 15. After a year, I left home and got
a job. I went through a bunch of different lives, something like
Picasso’s different periods. I studied Chinese literature
and art and philosophy and worked as a waitress. I was sculpting
and painting and writing poetry and hanging out in the cafes in
Greenwich Village. My friends were ballet dancers and painters
and musicians.
In 1975 I moved to the Bay Area from Manhattan to help my sister
raise her daughter, and entered a spiritual phase. I got into
Zen Buddhism for a while, then tried Hinduism with Indian guru
Sri Chinmoy. Then I got pulled into this bunch of born-again Christians
and married my first husband, with whom I had my three kids. By
now I think I’ve been successfully immunized against all
religions.
Q: That’s when you decided to go back to school?
K.S: Yes. I wanted to get a Ph.D. in the sciences. At
that point, I wanted to have something that was entirely mine,
something not dependent on a man. And of course I wanted interesting
and challenging work. I got that in spades!
Q: What was it like going back to college at age 32, and
why Santa Cruz?
K.S: Santa Cruz is a relaxed, funky little beach town
and the first time I drove through I felt at home. I knew I wanted
to study machine learning and artificial neural networks as ways
of modeling the brain. Santa Cruz had David Haussler, the leading
person working in these areas.
When I started school, I wanted to study human cognition. But
it became evident that, while computational neuroscience was interesting,
the time for computational molecular biology had arrived, and
the machine learning methods we were developing were powerful
tools in this field. It was thrilling to be a part of something
so real and important and to be able to contribute my own research
as an undergraduate. But to tell the truth, going into my senior
year I still wasn’t 100 percent sure that I wanted to stay
in computational biology.
Then David showed my work to [UCSC molecular biologist] Harry
Noller. At that time there was a big controversy over the branching
order in the tree of life. I had come up with clustering algorithms
that we applied to DNA sequences, and the phylogenetic tree I
constructed supported one side of the debate. Noller got interested
and suggested I take his course on the molecular biology of the
gene.
That course was like the last battle in The Chronicles of
Narnia, where these English children wander through the Kingdom
of Narnia and find a little hut in a glade. They go into the hut
and find an entire world there. When I took the class with Noller
I felt like I was opening the door to this hut, and that hut was
the cell. The world of the cell just opened up to me. These molecules,
proteins, were doing things! There is a whole dynamic life going
on within them, which is an absolute mystery. They were like living
things, not just inert, passive objects. It was a tremendously
exciting time. That feeling hasn’t stopped.
 |
"My
role at Berkeley is one of fostering an environment that has
a spirit of playfulness," says Sjölander. "It's
all about creativity and imagination."
PEG SKORPINSKI PHOTO |
Q: So you entered grad school at Santa Cruz knowing you
wanted to combine the computer modeling tools you’d developed
with the study of life and evolution?
K.S: The methods I applied toward protein phylogenic
tree construction can also be applied to species evolution. In
fact, a chapter of my Ph.D. thesis looked at the phylogenetic
relationship between humans, chimps, gorillas, orangutans, and
so on. But most of my work in phylogenetics has been identifying
protein superfamilies, which has a distinct set of problems.
Q: You use phylogenetic inference to identify how protein
superfamilies evolve novel functions and structures. Why is that
so tricky?
K.S: Phylogenetic or evolutionary inference has been
traditionally applied for speciation—figuring out how species
are related—and the biological processes underlying protein
superfamily evolution add a level of complexity. Also, proteins
have different forms at different times in their lifecycle, and
there are post-translational modifications. They interact with
partners and they have conformational changes. They form complexes,
they engage in the complex, and then they disengage from the complex.
There is a hubbub of activity in the cell. We’re just beginning
to understand how proteins do these things and under what circumstances.
When you look at these protein families at the molecular level,
it changes your view of life. You can line up related proteins
from bacteria, fungi, mammals, and plants in a multiple-sequence
alignment and see we all have a common ancestor.
Q: How did you come to have a joint appointment with the
Department of Plant and Microbial Biology?
K.S: I was here at Berkeley when I saw a program announcement
from the National Science Foundation for plant genome research.
When I give talks on how protein superfamilies evolve novel functions
and structures, one of my favorite examples is a little protein
superfamily that includes toxins made by scorpions as well as
plant and insect defensins [proteins that are part of the innate
immune arsenal]. When you look at this superfamily, you immediately
see that the common ancestor of plants, insects, and scorpions—some
primitive eukaryote—had some gene like these, which has
evolved to play different roles in the present day descendents.
In plants and insects, they’re part of the defense. In scorpions,
they’re part of the offense.
When I saw this announcement, I realized I could take the computational
methods I developed for protein superfamily analysis and apply
them to proteins involved in plant disease resistance. As it turns
out, many of the proteins involved in the mammalian innate immune
system are found in plants. And when we study the two systems
side by side, we get some interesting insights. I wanted to find
biologists to collaborate with, so I looked online and found a
Berkeley faculty member working in this area, Barbara Baker, a
plant and microbial biology professor. We met at a local cafe
and in the course of that first cappuccino decided to work together.
We’ve become great friends too. One thing led to another,
and I was invited to join PMB as an affiliated faculty member.
Q: We’re still in the midst of computational biology’s
big bang, but has the field said anything important yet about
the big theories or patterns of evolution?
K.S: One thing that has become very clear is that, although
evolution is described as largely tree-like, there is a lot of
horizontal gene transfer going on, particularly among bacteria.
Computational biology has really highlighted this.
Q: Speaking of horizontal transfers, when you are working
across disciplines, you rely a lot on collaborations.
K.S: Absolutely. My collaborations with experimental
biologists are crucial. They help ensure that we’re developing
methods that will address problems important to the biologists
and not just problems that are theoretically interesting. Working
with experimental biologists also helps hone our methods. We apply
a computational method to data that biologists are interested
in and get some predictions. Then the biologists do wet-bench
experiments and give us feedback. Then we look at the data together
and find out what parts we might not have been paying enough attention
to.
Q: Do you use these tools to corroborate and cross-check
the results that biologists are getting, or also to forge new
territory?
K.S: We sometimes find novel members of protein families
that the biologists didn’t know existed. They’ll tell
us that they’re looking for a protein with such and such
characteristics. We’ll find the proteins and they do experiments
on them and say whether we’re right or wrong. Or we predict
that a certain protein has a specific function based on our analysis.
Then they go and test that.
Q: Where do you see your work going now?
K.S: I’m now in a place where I can work on my
first love scientifically, trying to understand human cognition.
I hope to learn something about what we are as sentient animals
and how the protein families that confer that functional specificity
have evolved.
Q: You’ve said that David Haussler’s creativity
and playfulness, as well as his science, were inspiring and formative
for you.
K.S: Everyone in my lab knows how important it is to
me to be creative and playful but also to work hard. We have very
high standards of scientific excellence. You have to be rigorous
or it stops being fun.
Q: Your work has so many different facets. Is there a
central question in all of this diverse activity?
K.S: The central question is, ‘How does life evolve?’
Period. It’s the evolution of life and the understanding
of life. I address the question in a concrete way by studying
how protein superfamilies evolve novel functions and structures.
Q: Are you ever nostalgic for your bohemian years as an
artist?
K.S: When I was young I wanted more than anything to
be a great artist . . . and in a way that dream has come true.
I'm not the kind of artist I thought about being then; I'm a computational
artist. But I get to exercise my creativity, to use my mind and
imagination to engage the world and to explore it and to reflect
it. I just love that.
GORDY SLACK is an Oakland-based science writer
specializing in evolution and the environment. His work appears
in California Wild, Wired, Mother Jones, Bay Nature,
and Sierra.
|
