Solving problems requires coming up with ideas, but not all ideas are created equal. So how do you know if your idea is good?
Matthew Hirschey, Ph.D. associate professor of endocrinology, metabolism, and nutrition, had been thinking about this in relation to his research in gene pathways for some time. Then he came up with an (good) idea: Data-Driven Hypothesis (DDH) - a research tool that can predict pathways and functions for thousands of genes across the genome.
“Having data may help you formulate your idea better,”Hirschey said. “Data can help a person overcome any biases that may derail clear thinking.” The goal of DDH is to help researchers quickly and easily sift through volumes of publicly available biomedical data to come up with better hypotheses. A pre-print detailing this resource is available on BioRxiv.
Using a combination of computer programming, math, statistics and topical expertise along with the rapid adoption of open source science and data sharing allow scientists to access publicly available datasets and interrogate these data before performing any experiments.
“I hope this resource can be a clearinghouse, a place where students and fellows can go to explore genes, proteins and the relationships between them that might help crystallize ideas to then go test in the lab,” Hirschey said.
To use DDH, researchers just need to type in the gene they want to explore. If they don’t have a gene, they can click on “get lucky” and see a random gene to learn more about. That brings the user to a page of summary information. From here, users can decide if they want to look at cell dependencies, published papers, and graphs that show functional pathways. They can also download a report of all of the information of a given gene.
But the tool doesn’t just let users search for one gene at a time. They can actually search many genes using a pathway or custom gene list.
Initially, DDH was developed to help out the Hirschey lab. “I quickly realized that there are a lot of genes out there,” Hirschey said, “and this resource could be useful to other groups.” For the past year, he has been sending data to collaborators within Duke, but now researchers can access a web app.
Hirschey has plans to continue to build and improve the Data Driven Hypothesis tool. While he is already using three data sources, he hopes to increase that. So far, DDH is only using genetic information, but he has protein information that he also plans to include.
“Shortly after I made the website, I stopped using the programmatic interface to do my searches,” Hirschey said. “Instead, I started using the early version of the DDH website for all of my own searches because it’s just that easy.”