MIDAS Challenge Grant funds study of single cells


Research led by a team of scientists from across the University of Michigan makes the analysis of individual cells possible, unlocking potential advancements in biology research and a variety of other disciplines.

The study team, led by Jun Li and his colleagues at the Michigan Center for Single-Cell Genomic Analytics, used seed money from a Michigan Institute for Data Science Challenge Grant to generate more than $30 million in external funding and $7 million from the U-M Biosciences Initiative, to continue and expand their work.

“This poses a revolutionary shift in our ability to focus research on single cells,” said Li, a MIDAS affiliate and professor of human genetics, computational medicine and bioinformatics in the Medical School.

“Our project has had a strong pollination effect, spurring dozens of additional research teams to adopt single-cell methodologies, giving single-cell biology a transformative presence at U-M.”

Until now, the inability to examine individual cells meant questions pertaining to their functional differences were answered with crude measurements by researchers because even a small piece of tissue sample contained a heterogeneous mix of many cell types.

Being able to hone in on individual cells allows for a better understanding and identification of how the human body regulates growth, as well as the causes and efficacy of treatments for irregular growths, such as in cancer.

“So many biomedical discoveries are driven by technology,” Li said. “Many longstanding puzzles are suddenly solvable by these technological advances. MIDAS and U-M, as a whole, were nimble and resourceful in this space, treating this development as critical research infrastructure.”

These new advances in single-cell sequencing, however, come with challenges. Because a cell contains only a small amount of biological material, rare molecules are often missing in the data, which can lead to difficulties in assembling complete information.

This “sparse read counts data” presents both a challenge and an opportunity for U-M researchers. Data scientists and mathematicians are working in tandem with biological researchers to implement new methodologies in sparse data analysis.

“MIDAS Challenge Grants were established to bring data scientists and domain experts at the University of Michigan together to solve real-world problems, and to leverage data science services and infrastructure at the university,” said MIDAS Director H.V. Jagadish, the Bernard A. Galler Collegiate Professor of Electrical Engineering and Computer Science, professor of electrical engineering and computer science in the College of Engineering.

“By devoting resources to these existing strengths at U-M under a data-science focus, MIDAS’ goal was to produce extraordinary projects with major scientific impact.”

MIDAS is based in the Office of the Vice President for Research.

The methods being developed to infer missing information in the measured profile of a cell has broader potential impact outside of biomedical research. Social sciences, for example, could benefit from improvements in sparse data analysis.

“Political polling, by design, gathers opinion from a subset of individuals and has to adjust for trends to infer what whole-society opinion is,” Jagadish said. “There could also be applications for population data, as the U.S. Census is only taken every 10 years. The wider outcome of this project will produce general-purpose tools for many research areas.”

Li attributes the team’s wide impact and successes to the unique environment that U-M and MIDAS provide.

“U-M has been nimble in this space, we push to the very edge of what we can do because we have the resources and the talent to jump in early. … MIDAS’ initial funding was special as it connected (us) to an interdisciplinary ‘spider web,’ an intellectual gathering place. Having this network has been so effective for faculty and students.”


Leave a comment

Commenting is closed for this article. Please read our comment guidelines for more information.