Daniel Burkhardt has spent much of his academic career in laboratories of various kinds. But perhaps his biggest breakthrough came in a coffee shop in Vancouver.
During a break at a conference, he and other researchers using artificial intelligence to analyze single-cell data began commiserating in the cafe over a lack of standards — and standardized data sets — to objectively evaluate each other’s tools.
“By doing everything out in the open, you really bring people together to work on important problems, and then everyone can benefit.”
“We all realized that we’d like to try to compete with each other, and not spend all of our time thinking about how to compete,” said Burkhardt, who received a Ph.D. in genetics at Yale University and now works as a machine learning scientist at Cellarity.
Out of that conversation, Burkhardt co-founded Open Problems in Single-Cell Analysis, an international consortium of researchers developing a centralized, open-source framework for the evaluation of single-cell algorithms. The group developed a platform to evaluate algorithms on easily-accessible, standardized data sets, and began hosting competitions among researchers.
“What’s been really exciting about these competitions is that they’ve gone beyond just the computational biology community,” Burkhardt said. “We’ve actually gotten a lot of people from the machine learning community who’ve always been interested in single-cell, but didn’t know how to get started.”
More than 1,200 competitors are signed up for the latest competition. The broader goal, Burkhardt said, is to harness this collective brainpower to understand single-cell biology, and all of its complex interactions and connections, on a much deeper level — hopefully leading to the development of new therapeutics.
“By doing everything out in the open, you really bring people together to work on important problems, and then everyone can benefit,” Burkhardt said.
— Casey Ross