ExpressionBlast streamlines research for gene expression
ExpressionBlast, a new search engine developed by scientists at Carnegie Mellon University and Bar-Ilan University, helps scientists compare gene expression data with millions of other studies in order to find experiments with similar profiles.
Ziv Bar-Joseph, associate professor in Carnegie Mellon’s computational biology department, led the research team that published its work in the October issue of Nature Methods; Guy Zinman, Shoshana Naiman, Yariv Kanfi, and Haim Cohen co-authored the journal report. Multiple Carnegie Mellon employees and students also contributed toward developing the search engine.
Bar-Joseph’s research focused on creating a concise, systematic way to compare gene expression data.
“Everyone knows about DNA, the genetic material. We have exactly the same DNA in every cell of our body; however, all our cells are completely different in terms of what they do and look like [e.g., muscle cells vs. blood cells].... The reason that these cells are different is that different genes are being expressed,” said Zinman, now a research scientist in Carnegie Mellon’s Lane Center for Computational Biology.
Zinman explained that the idea for ExpressionBlast originated from a paper that explored longevity in mice. In this research, expression of a certain gene allowed male mice to live 15 percent longer than control male mice; this phenomenon only occurred in male mice. In order to make sense of the data, the researchers began to look for other experiments with similar results.
Currently, one of the main places to look for this information is Gene Expression Omnibus (GEO), the largest public database for expression data. GEO contains more than a million experiments, but does not provide an easy way for users to compare these experiments with their own results.
“As a research community, we had all this data, and it wasn’t being used as much as we would have liked. There was no way for you to compare your own results to the one million experiments that were out there,” Zinman said. “This is because every experiment uses a different machine, a different platform, a different animal, and is testing different conditions.”
ExpressionBlast automatically organizes data from genetic expression experiments on a weekly basis, so that they can be easily compared. “Everyone can now take their own experiment, put it on the web interface we developed, and find other experiments that have similar profiles,” Zinman said.
ExpressionBlast has many applications in and out of the realm of science. Drug companies, for example, could use ExpressionBlast to determine whether a potential chemical compound is affecting gene expression in a way shown to have the desired outcome in a previous experiment conducted by others, saving them time and money.
Zinman said that his work with ExpressionBlast was funded and greatly influenced by a program known as Innovation Corps (I-Corps), a program run by the National Science Foundation that focuses on giving entrepreneurial training to students and professors in order to foster commercialization of ideas that arise throughout research.
“It is a manner of thinking that is completely different from academia,” he explained. “I used this business thinking later on with my research, when looking for collaborators and research partners, and targeting my work towards more practical applications that the biological or medical communities are currently lacking.”
“The idea is to translate research to the marketplace,” Zinman continued. “There’s so much research being done at universities that can be applied to real world problems in a very short time, and many times the researchers don’t have the skill set to go outside of the comfort zone of their labs and look for customers and commercialization options.”
Aside from the possible long-term potential of ExpressionBlast, Zinman said that positive feedback for the tool is already surfacing. “I get about one to two emails per day. I’ve seen tweets from Cuba, I’ve gotten questions from people in India, and even thank you emails from researchers,” Zinman said. “If it wasn’t for the publication, it was for these emails. Just for them, it was worth it.”
Despite the success of ExpressionBlast, Zinman admits that there is still work to be done. “We still have many ideas that we would like to see implemented,” he said.
ExpressionBlast has made previously available gene expression data more useful, and has already begun to influence research worldwide. In the future, ExpressionBlast will undoubtedly affect the way in which gene expression data is analyzed and contribute to many aspects of the academic and commercial world.