Xiaohong "Jasmine" Zhou
Simon Tavaré
|
Cancer by the Numbers
Scientists Tap Math and Computers for Cancer Studies
By Eva Emerson
Computational biologists work in the world of numbers — populated by
equations, algorithms and binary code — situated in the biological
universe. Yet, in the future, their research will likely pay dividends
in a far less abstract realm: the doctor’s office.
USC College’s Simon Tavaré presents a prime example. A professor of
biological sciences, mathematics and preventive medicine, Tavaré holds
a doctorate in statistics. He’s also part of a team working toward a
future in which doctors may genetically profile a patient’s tumor to
guide treatment decisions.
His colleague Xianghong “Jasmine” Zhou studied biochemistry and
computer science and trained in bioinformatics at Harvard. Now an
assistant professor of biological sciences, Zhou creates data mining
software that could lead to new insights into cancer.
“It is clear that the mathematical sciences — in particular
probability, statistics and computer science — have already played a
major role in recent successes in molecular biology,” said Tavaré, the
George and Louise Kawamoto Chair in Biological Sciences. “This is sure
to continue as new experimental techniques generate new data that in
turn need new mathematics for their synthesis and interpretation.”
DNA microarray technology, also called gene or DNA chips, has emerged
as a powerful and widely used method to probe the complex workings of
the cell. DNA chips offer a way to visualize gene activity across the
genome in a single scan. What’s more, chips can reveal which genes step
up or slow down in diseases such as cancer and diabetes. They also
provide an alternative way to pinpoint novel disease-linked genes.
The size of a postage stamp, a single microarray can hold DNA fragments
from all 30,000 human genes. The DNA fragments are embedded in a glass
slide or silicon chip. In an experiment, genetic material from a sample
is washed over the array. Matches between the sample and DNA “light up”
the chip, producing a pattern of brightly colored dots. Computers read
the chips optically, collecting information on intensity and color. The
data then undergo a series of mathematical and computational analyses.
Comparing gene activity in two different cell types, such as healthy
cells and tumor cells, can reveal telltale patterns called genetic
signatures of cancer. Scientists have used microarrays to find genes
important in prostate and breast cancers. Others revealed a genetic
signature associated with aggressive breast cancer tumors likely to
spread to the lungs.
But microarrays face a number of hurdles. Each experiment produces a
flood of data, which can overwhelm researchers. The data are noisy, and
results of experiments can be difficult to compare directly, even when
the same kind of DNA chip system or platform has been used. Data
generated from different platforms have been near impossible to compare.
Microarrays and Cancer
Tavaré’s research group has addressed some of these issues in its work
designing and analyzing microarray experiments in cancer genomics.
“Microarray technology is used in many different aspects of cancer
research,” said Tavaré. “One way is to look at the patterns of gene
expression in a set of individuals with a particular cancer in order to
predict survival or response to chemotherapy. This technique may
provide a more reliable way to classify tumors than classic methods of
pathology.”
With colleagues at Cambridge University, Tavaré studies large-scale
genetic alterations in tumor cells. Specifically, the team looks at
gross changes in DNA regions known to contain tumor suppressor genes
and oncogenes — families of genes often altered in tumor cells. Mutated
tumor suppressor genes fail to halt cell growth, and damaged oncogenes,
turned on inappropriately, trigger uncontrolled growth.
Tavaré and his colleagues also are following women with breast cancer
tumors categorized as low-, medium- or high-grade by pathologists and
then characterized by microarray analysis. They are studying the women
as they undergo treatment in an effort to identify the most effective
therapies for each tumor type.
The eventual goal is “when a new person with a breast tumor comes
along, you could test them and [based on the results] predict which
chemotherapy will work best for them,” he said.
Tavaré is no stranger to cancer research. He has long collaborated with
USC molecular pathologist Darryl Shibata to understand how, where and
when healthy cells in the colon mutate into malignant cancer cells.
Tavaré has tapped a technique called genetic evolutionary analysis,
which tracks accumulated changes in the DNA sequence, to trace the
lineage of tumor cells. He has used similar “molecular clock”
approaches to infer the evolutionary distance between species and
populations.
Solving the Cross-Platform Problem
While Tavaré focuses on fundamental issues of reading and interpreting
microarrays, computational biologist Jasmine Zhou concentrates on
comparing results from the enormous number of completed microarray
experiments recorded in public databases.
“Microarray data have been flooding in,” said Zhou. But “due to the
“cross-platform” problem, few have actually made use of what’s in the
databases. The ability to compare data from different research groups
would be a boon to cancer researchers.”
Zhou recently unveiled a new software program, the Integrative Array
Analyzer, designed to help scientists do just that. Mining microarray
databases and integrating the data collected thus far would increase
confidence in the findings, she said.
The program might also help reveal new genetic signatures common to
different cancers, a possibility that interests Zhou. She anticipates
using the software to identify interactions between genes that may be
associated with cancer, but when looked at a single gene at a time, may
not appear important.
“I want to know how genes interact, what they do and how they are
regulated,” she said. DNA microarray data will be crucial to that aim.
“It gives you a snapshot of a particular moment in a cell,” Zhou said.
“It shows you what’s happening across the entire genome. By integrating
multiple data sets, we can begin to focus on the discovery of networks
of genes, and how they are differentially regulated in cancer tissues.”
“By itself, a list of genes turned up or down in cancer cells will not
tell us all we need to know to understand complex diseases such as
cancer. Often, it is the interaction of many genes that causes the
differences in tumor cells most critical in medicine, such as the
severity of disease.”
|
|
|
