X
Innovation

Cancer? Google it

How search engines could change the way researchers look for cancer markers.
Written by Rose Eveleth, Contributing Editor

There are about 20,000 proteins that might be involved in pancreatic cancer. How do you know which ones are the most important? Well, Google might know. Or at least, they might know how to figure out.

Scientists at the Dresden University of Technology in Germany recently used a modified version of Google's PageRank algorithm to go through those proteins and figure out which ones might be more important than others. A study on their work was released today, in the journal PLoS Computational Biology.

When looking for cancers, much attention has been paid to something called "biomarkers." Biomarkers are basically just things that are made by cancer cells. The idea is that if you know the compounds a cancer cell makes, you can look for those and find the individual cells rather than having to find a whole tumor. It could help doctors find cancers before they grow. But cancer cells make a whole lot of biomarkers, some more common than others. And finding biomarkers is hard and expensive, and no two studies seem to find the same ones.

Enter: search algorithms. The basic premise behind page rank is that Google takes into account what's on the web page, and how the page is linked to other pages. The more times a page is linked to, the more authoritative it is, and the higher it ranks. Researchers thought they could do the same thing with biomarkers. The more often these proteins were connected to one another and to different things happening in your body, the higher they ranked.

"Once we added the network information in our analysis, our biomarkers became more reproducible," Christof Winter, the paper's first author, said in the press release.

As with everything in science, there is still a ways to go. Using the search algorithm might help figure out which biomarkers are the most common or the most linked, but it remains to be seen whether those are also the most useful for diagnosis. But at least now scientists know how to look.

Via: Eurekalert

Image: Danard Vicente, Flickr

This post was originally published on Smartplanet.com

Editorial standards