Unraveling the Complex Functions of Proteins

Automated Function Prediction conference launches a new field

At the birth of a new field, a conference can act as a midwife, making sure the infant enters the world smoothly. Such was the case for the Automated Function Prediction (AFP) conference held at the University of California, San Diego, at the end of August 2006. The field, in which researchers try to computationally determine a protein’s function, appears to have arrived in good health.


AFP conference participant Jurgen Schulze, PhD, receiving an immersive experience of protein functional sites in the virtual reality cave at California Institute for Telecommunications and Information Technology. The software supports collaborative viewing of proteins at multiple sites on the internet. Courtesy of Jurgen P. Schulze.“I feel we went quite far in creating an identity for this field,” says Adam Godzik, PhD, program director in bioinformatics and systems biology for the Burnham Institute for Medical Research in La Jolla, California. The conference drew independent researchers who worked together for the first time to establish a common language, Godzik says, as well as a compilation of available methods and a way to evaluate the success of various techniques.


For decades, biochemists have tried to solve the riddle of what individual proteins do. They’ve done this in the lab, painstakingly slowly. But scientists now sequence genomes far faster than they can assign functions to the corresponding proteins. When there were only a few sequenced genomes, this problem was relatively small in scale and seemed manageable. But, says Iddo Friedberg, PhD, the conference organizer and a postdoctoral associate in Godzik’s lab, “By sheer scale the problem has changed.”


Now, using computational methods, researchers can tease through to answers much more quickly. At the AFP conference, those adept at the process charted the future of this discipline.


Their approach weaves together methods biologists have used for years. They include physical analysis of a protein’s structure; genomic context, in which researchers compare a protein’s position in a gene to those of similar genes elsewhere with known functions; and what Friedberg calls “guilt by association,” or information gleaned in the laboratory about what the protein does when a cell undergoes a particular process. But because they’re using computers and high-level algorithms, AFP researchers can now analyze more information faster and come to more robust conclusions about the workings of previously mysterious proteins. Visit http://BioFunctionPrediction.org for more information.


Godzik was pleased with the progress the conference made in crystallizing the field. “Until now,” he says, “automated function prediction has been a black art.”



Post new comment

The content of this field is kept private and will not be shown publicly.
This question is for testing whether you are a human visitor and to prevent automated spam submissions.
Enter the characters shown in the image.