AI to help researchers see the bigger picture in cell biology

Source Domain: news.mit.edu

Researchers from MIT, the Broad Institute of MIT and Harvard, and ETH Zurich have developed an AI-driven framework to analyze complex cellular information across various measurement modalities, such as gene expression and protein measurement.
The framework identifies both shared and unique data from different measurement types to provide a more comprehensive view of a cell’s state.
This approach allows researchers to better understand intracellular interactions and mechanisms for various diseases like cancer, Alzheimer’s, and diabetes.
Traditional machine-learning methods blend data from different modalities, making it hard to discern their source, whereas the new method distinguishes data in a manner similar to a Venn diagram.
The researchers applied the framework to synthetic and real-world single-cell datasets, demonstrating its ability to automatically differentiate between shared and modality-specific data.
The method helps to determine which measurement technique should be used to detect specific cellular markers, potentially streamlining the research process.
The team aims to make the model more interpretable and expand its application to more clinical and biological questions in the future.

AI to help researchers see the bigger picture in cell biology | MIT News