IBM vs. a grain of rice

By turning to idle computers scattered around the world, IBM and computational biologists at the University of Washington hope to speed up the development of the perfect strain of rice. The hunt for super rice is part of the World Community Grid, which was launched in 2004 and has grown to be the third most powerful supercomputer in the world, according to BusinessWeek. The new rice genome work is one of a handful of distributed supercomputing projects that IBM’s Grid is orchestrating—their first and largest project used ordinary PCs distributed around the world to predict how proteins fold, and ongoing research targets developing better models of climate change in Africa, understanding the role of proteins in fighting cancer, and searching for new drug candidates.

People join as volunteers and download software that, when their machines are not being used, communicates with the Grid’s servers to request data on a project. The software uses prediction algorithms to perform computations on that data, such as to guess at the structures of proteins in major rice strains, and the role those proteins play in interacting with other genetic material in rice and nutrients.

The first step is to better understand the shapes of rice proteins, then researchers can more easily decipher their functions and how they interact to produce particular characteristics of rice—for example, to produce greater crop yields and disease resistance.

Then once the researchers have narrowed down the selection process of desirable proteins to produce optimal strains, these new hybrids will move from the lab to be tested in greenhouses.

Ram Samudrala, the project lead, expects to get new rice strains to farmers in the next five years—far faster than if they used only the school’s supercomputing resources.

Early distributed computing projects included SETI@home, which searched for extraterrestrial intelligence, and folding@home, which aimed to understand how human proteins form their complex structures. For more background on volunteer computing in science and the people who developed it, check out BOINC, the Berkeley Open Infrastructure for Network Computing.