Grid computing is a source of enormous untapped power that could
revolutionize life sciences research, according to speakers at a Web
conference hosted by United Devices.
"Internal Grids are an enormous untapped resource," said Scott Kahn of
Accelrys. "It's already in-house, it's already running, and it's already
being maintained. It's rare that in a day, you can change your
processing power by a thousand times."
Stuart Henderson of PwC Consulting said life sciences' evolving R&D
model will be a "big driver" of Grid computing's growth. The emerging
"learn and confirm" model will stand the old R&D model on its head by
placing in silico research first and research on patients and in
laboratories second.
The new model could give Big Pharma companies an advantage over their
biotech rivals, Henderson said, because the big drug firms have many
more machines at their disposal. "Big Pharma could use its greater
compute power to widen the lead over biotech firms," Henderson said.
Graham Richards, chairman of Oxford University's chemistry department,
called Grid computing's contribution to research "one of the biggest
steps forward in my entire lifetime."
The explosion of data from the decoding of the human genome has created
a need for enormous compute power, Richards said, as the "few hundred
thousand genes" are broken down still further into proteins and
molecules. Oxford's massive global volunteer cancer screening effort, in
concert with United Devices and Intel, is in the process of in silico
screening of 3.5 billion drug-like molecules for their ability to bind
with proteins. The year-old project has 1.5 million PCs in 215 countries
donating 100,000 years of CPU time, including three PCs in the Antarctic
and even some in Vatican City.
Kahn said Grid computing solves two needs that life sciences firms have:
to improve productivity, and to crunch the huge explosion of data in
recent years. "There is more raw data, and less time to analyze it," he
said.
Kahn said the least intrusive Grid applications are those with a small
memory footprint that don't write a lot to disk. He cited LigandFit as
an application that "runs as fast or faster on a Grid desktop as it does
on a Unix server."
Events
IE 7
Firefox 2.0
