Chan Zuckerberg Initiative grant to the Glastonbury Group

The Glastonbury Group is among the recipients of the Data Insights Cycle 3 awards. The aim of the grant is to develop a machine learning model that identifies disease-relevant cell subpopulations whilst predicting a phenotype/disease of interest from large-scale single-cell RNA-seq data.

The Chan Zuckerberg Initiative (CZI) recently invited a third cycle of researchers to apply for 18-month long projects focused on supporting computational experts in advancing the tools and resources that make it possible to gain greater insights into health and disease from single-cell biology datasets.

The awardees of the third cycle of the project were announced on 23^rd July. The 17 awarded projects include the proposal submitted by Craig Glastonbury, Group Leader at Human Technopole, in collaboration with Nicole Soranzo, Senior Group Leader at the Wellcome Sanger Institute and Head of the Genomics Research Centre – Population and Medical Genomics Programme at Human Technopole.

The project, titled “Phenotypic prediction from population-scale single-cell RNA-seq”, aims to develop a framework that enables the prediction of broad disease, trait and biomarker from single-cell data while simultaneously identifying relevant implicated cell types. 

The availability of biobank-scale single-cell studies coupled with foundation models trained on cell atlases presents an incredible opportunity to develop frameworks that enable broad disease prediction and prognostication and a more complete understanding of disease heterogeneity at the level of single-cell subtypes and cell states.

To enable this, the Glastonbury Group will develop a framework that at its core utilises advances in Multiple Instance Learning (MIL). With this approach, they will probabilistically classify donor single-cell profiles according to their disease status or biomarker abundance and simultaneously identify implicated models and MIL methodologies, marking a paradigm shift in enabling disease prediction, prognostication, and single-cell understanding of disease heterogeneity. Specifically, the Glastonbury Group will leverage single-cell RNA-seq profiles collected in 6.500 densely phenotyped UK biobank donors. They will produce an atlas characterizing how cell types/activation states are implicated in a specific disease, their progression and a range of molecular and metabolic biomarkers.

“The framework of utilizing single-cell data for accurate disease prediction and prognostication will improve the current mechanistic understanding of how diseases, traits and biomarkers play out at the level of individual cells and pave the way for a more complete understanding of disease heterogeneity in a population” underlined Craig Glastonbury.