A new artificial intelligence (AI) tool developed by researchers at the Wellcome Sanger Institute, the Institute of AI for Health at Helmholtz Munich, the University of Würzburg, and collaborators has shown it is capable of rapidly analyzing and interpreting millions of cells from a patient sample, accurately predicting molecule changes in the tissue. Dubbed the NicheCompass and developed as part of the Wellcome Sanger’s Human Cell Atlas Initiative, the tool creates a visual database that combines spatial genomic data on cell types, where they are located, and how the cells communicate with each other.
Details of NicheCompass, published in Nature Genetics, show it has the potential to determine which personalized treatments could be most effective for diseases such as cancer.
“NicheCompass is a significant leap in this field, leveraging the power of AI but also offering interpretability, allowing researchers and clinicians to ask questions about their data and better understand and treat diseases,” said first author Sebastian Birk, a data scientist at the Institute of AI for Health, Helmholtz Munich and the Wellcome Sanger Institute.
NicheCompass is a deep-learning AI model that is built to recognize and interpret cell-to-cell communication. It is able to measure and interpret a range of cellular data to show how cells communicate through their networks and then aligns them with similar networks of cells, effectively “neighborhoods” of tissues with shared features. This ability to analyze cellular communications is crucial because every cell in the human body communicates with its environment and is involved in a larger network of interactions.
In the study, NicheCompass was used to analyze lung cancer and breast cancer tissues, demonstrating its ability to identify differences in how immune cells communicate with tumor cells. This analysis not only uncovered new information about the mechanisms of cancer, but also detailed how one patient’s cancer interacted with the immune system in a different way. In addition, the AI model can distinguish subtle differences between individual patients’ cancer microenvironments, helping predict how each patient might respond to treatment.
“Using NicheCompass, we were able to see the differences in how immune cells interact with lung cancer tumors in patients,” said co-senior author Carlos Talavera-López, PhD, from the University of Würzburg.
NicheCompass, the researchers said, is the first tool of its kind to use a deep-learning approach to analyze cell-to-cell communication and interpret these “neighborhoods,” which can be critical to understanding how cancer progresses. By doing so, it generates a map of how different cells within a tissue interact, which can be used to identify potential targets for drug development or to guide clinical treatment decisions.
The team also tested NicheCompass on a variety of spatial genomics datasets, including data from mouse brain tissue and embryonic development. It was able to process over 8.4 million cells in a mouse brain atlas, correctly identifying brain sections and creating a detailed visual resource of the organ.
Because of its versatility, NicheCompass’s applications extend beyond cancer. Having a tool to analyze the spatial atlases of various organs, including brain and lung tissues, provides an unprecedented tool to help better understand disease mechanisms at a molecular level. Further, patient-specific data can also be incorporated to allow clinicians to receive in-depth information about an individual’s condition, allowing for more tailored treatments.
Researchers are also optimistic that NicheCompass can be adapted for use in multi-omics studies, offering insights into tissue architecture, immune responses, and the complex gene programs that regulate cellular behavior.
This post was originally published on this site be sure to check out more of their content
