STANFORD--Researchers at the Stanford Human Genome Center have developed
a powerful new computer program that can map thousands of genetic markers
at once. Using this program, called Mapper, scientists have obtained one
of the most accurate views yet of the human genome as a whole.
"These maps have better resolution in local areas than the other available
whole genome maps," said Dr. David Cox, professor of genetics and co-director
of the Stanford Human Genome Center. "The others can tell you what
block the house is on, but we can give the exact address."
Mapper produces high-quality genetic maps that will dramatically advance
the efforts of the Human Genome Project, whose goal is to map and sequence
the genetic blueprint that contains all of the instructions about how to
make a human body, said Richard Myers, professor of genetics and director
of the Stanford Human Genome Center.
"Previously, it would take months for us to build a map," said
Myers. "Now we can rebuild it every day, including new information
as it is gathered. Our maps are now much more accurate because we can map
everything at once."
The Stanford team that developed Mapper will present the new method, this
week, along with the maps it has generated, at the Genome Mapping and Sequencing
meeting at Cold Spring Harbor, N.Y., May 8-12.
To get an idea of the challenge confronting gene mappers, imagine constructing
a map of the world knowing only that Palo Alto is close to San Francisco,
Paris and Frankfurt are both in Europe, and Johannesburg and Nairobi are
on the same continent.
Gene mappers use short, distinctive genetic markers found in DNA as signposts
in the human genome. Markers that are close together remain on the same
piece of DNA when scientists cut the long strands into small pieces. Using
statistics to count how often particular pairs of markers remain together,
computers can measure the distances between them and can determine their
order on the DNA.
Until now, generally available computer programs have had to test every
possible combination of locations for each set of these DNA signposts to
figure out which one was correct. This method takes a large computer a relatively
short amount of time if processing only 15 or 20 markers. However, each
of the human chromosomes contains 500 to 1,000 of these markers -- far too
many to map all at once using available computer programs.
To solve this problem, computer scientists at the Stanford Human Genome
Center applied a well-known computer trick: They instructed the computer
to ignore some unlikely possibilities instead of demanding that every possibility
be checked.
"This technique is not exact or exhaustive -- and that's why it works.
The computers don't try everything -- that would take forever -- but they
come up with pretty good answers," said Kathleen McKusick, computer
scientist at the Stanford Human Genome Center. She and Cox played the major
roles in developing the new computer program.
To create Mapper, the scientists started with a very thorough program developed
by Michael Boehnke at the University of Michigan. But that program eventually
fell by the wayside.
"We built a car around an engine, and then changed the engine,"
said McKusick.
McKusick and colleagues modified the original program in steps, gradually
replacing the very precise, but time-consuming and expensive parts. At every
point, they evaluated its success to make sure they were not sacrificing
accuracy.
"We started out using the off-the-shelf thing, then tried to scale
it up. We found that we were bumping up against its limitations -- then
we'd change it. Eventually, the original program just dropped out of the
picture," said McKusick.
So far Mapper is performing well, she said. The researchers have tested
Mapper's accuracy by comparing its results with genetic maps obtained by
other scientific methods.
The Stanford team believes that Mapper will be invaluable as scientists
continue toward their goal of determining the entire sequence of genetic
letters that make up the human genome.
"The maps we've made using this program are among the highest-resolution
maps that have been made to date, but they're not detailed enough to allow
sequencing the whole genome. Next, we'll do more of the same but with five
times the number of markers so that we can use the maps for sequencing,"
said Myers.
The coming maps will also help researchers find the particular DNA markers
that are used to diagnose disease. "The more detailed maps are essential
in identifying disease genes, especially for diseases that are inherited
in a complex manner and aren't due to a defect in a single gene," said
Myers.