Nov

6

Metagenomics and the Computing Challenges of Microbial Communities

Filed Under big science, research horizons | Comments

Why should you care about microbial communities?
Except for viruses, they are the most abundant life on Earth and have an
overwhelming effect on our environment and our lives. Consider that about
half the carbon dioxide on Earth is processed through microbes that live in
the oceans. Then consider that the most modern climate models of ocean life
include just five organisms. This is despite recent findings that point to
thousands of oceanic species, which do many different things and presumably
influence our climate.

Metagenomics is a relatively new field that seeks to understand the
structure and function of the shockingly large number of microorganisms on
our planet.  New technologies permit us to now sequence samples taken from
their environment rather than only those that are cultivated in the lab. For
example, Craig Ventner’s Global Ocean Sampling Expedition has collected water throughout the world’s oceans, captured organisms, and sequenced their DNA. In the initial pilot study alone, nearly 150 new bacteria were discovered through this process.

The science and computing challenges are huge. A single gram of soil
contains approximately one trillion base pairs of DNA. Scientists at the National Institutes of Health recently compared over 100,000 bacterial gene sequences on the human skin and discovered a far larger number of different bacteria living on human skin than had been previously known (Science, May 28, 2009). Sequencing and making sense of these data introduces new computational problems, not merely slight extensions of existing ones.

The potential impacts of understanding these data are huge as well. In the
case of soil, microbial communities have an impact on carbon sequestration
and understanding them may help us with cleaning toxic waste. In our bodies,
microbial cells are estimated to outnumber our human cells by a factor of
ten to one and are important in protecting our skin, digestion, and much
more. Understanding these large microbial communities is therefore likely to
have a positive impact on human health. The NIH has launched the Human
Microbiome Project to support work in this field.

Complete DNA sequences of thousands of organisms are piling up in databases
because of the efficiency of DNA sequencing technologies. Most of this
remains unanalyzed for several reasons. We don’t yet know the right
biological questions to ask. We don’t have all the clever programs that
would actually ask these questions of the computer. And there is now so much
data that many questions totally overwhelm even existing high performance
computers.

Among the computational challenges in this field are the design of new
algorithms and cloud computing technologies. In the National Academies of
Science publication “The New Science of Metagenomics: Revealing the Secrets
of our Microbial Planet”, the authors conclude “What then, will metagenomics
have become, in 20 years? We believe that it too will be a concept-driven
computational science… We can expect, in 20 years, enormous advances on
three fronts – technical, computational, and biological – as well as a host
of specific applications.”

We encourage our community to explore and engage in this and other emerging
fields at the crossroads of biology and computation. This is one of the
exciting areas for 21st century computing.

Contributed by Bill Feiereisen with assistance from Ran Libeskind-Hadas

Sep

25

Inducing Innovation with Prizes

Filed Under big science, policy | Comments

The awarding of the $1 million Netflix Prize this week reopens an interesting bigger question:  Are prizes a viable mechanism for encouraging research in the computing fields?  From Netflix’s perspective, the answer is almost certainly yes.  Netflix CEO Reed Hastings is quoted telling the New York Times (probably tongue-in-cheek) “You’re getting Ph.D.’s for a dollar an hour.”

Could prizes be useful to the broader computing community in advancing research?  The Clay Mathematics Institute established the Millenium Prizes in 2000, offering $1 million for the solutions to each of seven famous open problems, including the question of whether P=NP.  It’s hard to imagine that many researchers have decided to shape their research agendas based on the existence of this prize.  On the other hand, Wolfram Research sponsored a $25,000 prize, with a blue ribbon prize committee, to determine if a specific small (2 states and 3 symbols) Turing Machine is universal. The problem was solved (in the affirmative) in 2007 by a 20-year-old from Birmingham, England.

There is a rich history of prizes for technical innovation.  In the early 18th century, the British Parliament offered the Longitude Prize for a practical method of precisely determining a ship’s longitude, with different monetary amounts depending on the accuracy of the instrument.  The rules were changed during the course of the competition and the prize was never awarded.

More recently, there have been numerous technical prizes such as the $10 million Ansari X PRIZE for carrying three people to 100 kilometers above the earth’s surface.  Following on the success of the Ansari Prize, The X PRIZE Foundation has established several other major prizes for specific achievements that have “the potential to benefit humanity”.

Are there some major problems in computer science that could be incentivized by prizes – financial or otherwise?  What are the potential benefits and risks of this approach?  We’re eager to hear your thoughts.

Some good additional readings include the following:

Aug

23

Cross-layer Reliability Visioning Progress

Filed Under big science, research horizons, workshop reports | Comments

The Cross-layer Reliability Visioning Study Group met July 8-9, 2009 in Los Alamos, NM.  This was the second of three scheduled meetings focused on how to address the growing challenges imposed by changes in device technology, system sizes, and application requirements.  A major goal of the Visioning process is to reach some consensus on how to achieve reliable computing using unpredictable components across different layers that dictate system reliability (i.e., device technology, design, architecture, software).  While the first meeting focused on defining the multi-dimensional cross-layer reliability design space, including both theoretical and practical aspects of the problem, the second meeting focused on considering cross-layer reliability from different application domains (e.g., consumer electronics, space/avionics, etc.).  The attendees were divided into visioning groups to target these individual domains.  Other visioning groups focused on developing common reliability metrics to address the cross-layer abstraction issue and addressing the technology reliability roadmap.  A number of common themes across the individual domains emerged, which will help to build consensus across the community as a research agenda is defined.

The third meeting will likely be scheduled for late October, though draft vision/consensus documents are being crafted before this next meeting.  The meeting will held at IBM in Austin, Texas, and will engage leaders from funding agencies as part of the program.

Next Page →