Skip Over Navigation Links

Web Exclusives: Movies

Computational Tool for Combing Through 'Hairballs' of Data
Emily Carlson
Posted May 27, 2010

Prettier than a hairball, but just as intricate. This diagram of the human interactome shows physical (green edges) and genetic (white dots) interactions. Credit: Keiichiro Ono
Prettier than a hairball, but just as intricate. This diagram of the human interactome shows physical (green edges) and genetic (white dots) interactions. Credit: Keiichiro Ono
Click for larger image.

In the bustling shopping district of Tokyo, computer scientists spilled into a large room inside Japan's flagship Apple store. They were there not to try out the latest gadgets, but to see how their colleagues visualize scientific data, particularly from the life sciences.

While the presentations focused mainly on Japanese-based visualization tools, such as E-Cell 3D Exit icon and KEGG Atlas Exit icon, Keiichiro Ono's talk demonstrated a popular U.S. project he works on called Cytoscape Exit icon.

As the name implies, Cytoscape integrates all known information about a cell, including its gene activity, biochemistry and protein interactions, into a single image. During his demo, Ono showed how users can connect to or upload data to generate a visual that they can customize by color, layout and even language—all in about an hour.

Cytoscape started in 2002 with the main goal of helping biologists make sense of the data streaming from genome sequencing projects. Its creator, Trey Ideker, explains, "The amount of data you can collect about a cell is exponentially increasing. One of the big questions that has arisen is how do you integrate it all together."

Ideker, a systems biologist at the University of California, San Diego, wanted to develop a freely available computer program that could help biologists piece together all this molecular information into a more complete picture of a cell. Now, more than 10,000 researchers (not all biologists) use the tool each month to produce pictures that help them explore and explain their data.

While the pictures are technically called network wiring diagrams, Ideker somewhat jokingly refers to them as "hairballs." The entire image represents the cell's molecular network, with nodes showing individual molecules (like proteins) and the lines showing the relationships between them. Because a cell may have thousands of interacting molecules, Ideker says: "It can be hard to pull out which ones are important."

Cytoscape includes many options that let researchers customize colors, layouts and even languages. Credit: Keiichiro Ono
Cytoscape includes many options that let researchers customize colors, layouts and even languages. Credit: Keiichiro Ono
Click for larger image.

To help tease them apart, Cytoscape includes computational tools that allow scientists to study the interactions most relevant to their own research projects. If they're interested in evolution, they may compare networks from different species—mice and humans, for example—to find areas of the network that are similar and thus conserved over time. If they're interested in a particular health disorder, they may search for network traits unique to diseased cells.

Ideker has used Cytoscape in his own research, which is partly focused on developing computational models of disease. Recently, he teamed with a cancer researcher to diagram molecular data from people with chronic lymphocytic leukemia (CLL), the most common form of leukemia. The effort revealed particular protein interactions that seemed impaired. These abnormal network activities could serve as biological markers for detecting CLL.

With potential medical applications like this, Ideker says Cytoscape may prove as powerful for clinicians as it has for biologists.

"In just a few years, we will be able to get our genomes sequenced at the doctor's office, and the only way to make sense of that information will be in the context of molecular networks," predicts Ideker. "My vision is that we'll see Cytoscape in the doctor's office as a tool for assessing patient health."

Learn about related research

This page last reviewed on April 22, 2011