Software tackles protein pathways

When biologists want to compare different sequences of DNA or protein, it’s as simple as plugging the information into a browser and pressing enter. Within 15 seconds, an online software tool contrasts one sequence of DNA with up to 18 million others catalogued in public databases. Now, a software tool developed by Whitehead Institute scientists promises to apply this same computational muscle to the far more intricate world of protein interaction networks, giving researchers a new view of the complexities of cellular life.

DNA sequencing technologies allow scientists to easily identify genes and their nucleotide building blocks — linear strings of information represented by the letters A, C, T and G. The wide accessibility of these technologies has enabled both companies and academic labs to assemble huge libraries of genomic information. Computer engineers, in turn, have helped scientists navigate these oceans of data through tools such as BLAST, the primary software platform that scientists use to compare protein and DNA sequences. However, many researchers believe that the next phase of genomics research will be to map out interaction networks — the cell’s internal wiring system through which genes and proteins communicate.

“The 80s and 90s were about sequences,” says Trey Ideker, a former Whitehead Fellow who recently was named an assistant professor of bioengineering at University of California, San Diego. “Now we’re starting to see newer types of technologies — like microarrays — that allow us to look at how a cell, in its entirety, responds to drugs and other kinds of stimuli. These technologies will revolutionize biology.” Already, researchers like Whitehead’s Rick Young are beginning to assemble libraries of cellular network pathway maps using microarrays.

“But there’s a problem that’s not yet addressed,” says Whitehead Fellow Brent Stockwell. “What if I’ve identified a whole protein interaction network in one type of organism and I want to see if a similar network exists in other species?” Until recently, there was no way to do this. It’s a need that Stockwell and Ideker hope their new software tool, called PathBLAST, will meet.

At the core of PathBLAST is a program that can represent these interaction networks mathematically. The program is based on algorithms that scientists use to represent chemical structures. “An interaction network, in its form, is essentially like a chemical structure,” says Stockwell, “and fortunately there are already a great set of tools for representing chemical structures.” Developed by Brian Kelley, a software engineer in Stockwell’s lab, this algorithm translates all the information from an entire interaction network into a linear code. Using an interface developed by Whitehead’s Biocomputing group, PathBLAST can rapidly compare interaction networks from different organisms.

In research published this week in the online edition of the Proceedings of the National Academy of Sciences, Ideker and Stockwell took the entire genomes from the yeast S. cerevisiae and the bacterium H. pylori and compared the interaction networks in both organisms. The software crunched the numbers and displayed the results in seconds. The turnaround time was impressive considering the scope of the effort: the bacterium contained 1,465 interactions among 732 proteins: the yeast contained 14,489 interactions among 4,688 proteins.

The study revealed that one pathway critical in catalyzing DNA replication and another one instrumental in protein degradation were conserved in both organisms as a single network. “What was surprising was that there was one network, not two,” Ideker says. “So now the question is, ’What’s the attraction between these two complexes?’”

At the moment, there’s no clear answer. But as labs continue to do these types of experiments, there eventually could be a huge payoff in comparing such things as viral networks to human networks, possibly allowing drug companies to develop products that target cellular pathways unique to viruses.

As for other applications, it’s still too early to tell, Ideker says. “It’s like asking in 1985, ’What’s the impact of gene sequencing going to be?’ We’re trying to get the basic mechanisms in place to eventually do these kinds of comparisons.”