Cross-domain information fusion for enhanced cell population delineation in single-cell spatial-omics data

Bokai Zhu ^1,2,3, Sheng Gao ⁴, Shuxiao Chen ⁴

¹Ragon Institute of MGH, MIT, and Harvard, Cambridge, MA, USA.
²Broad Institute of MIT and Harvard, Cambridge, MA, USA.
³Massachusetts Institute of Technology, Cambridge, MA, USA.
⁴Department of Statistics and Data Science, The Wharton School, University of Pennsylvania, PA, United States.
⁵Center for Virology and Vaccine Research, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA.
⁶Dana-Farber Cancer Institute, Boston, MA, USA.
⁷Center of Hepato-Pancreato-Biliary Surgery, The First Affiliated Hospital of Sun Yat-sen University, Guangzhou, Guangdong, People's Republic of China.
⁸Division of Genetics and Genomics, Boston Children's Hospital, Harvard Medical School, Boston, MA, United States.
⁹Program in Biological and Biomedical Sciences, Harvard Medical School, Boston, MA, USA.
¹⁰Division of Gastroenterology/Liver Center, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA, USA.
¹¹Department of Pathology, Brigham and Women's Hospital, Harvard Medical School, Boston, MA, United States.
¹²Institute for Medical Engineering and Science, Massachusetts Institute of Technology, Cambridge, MA, USA.
¹³Department of Chemistry, Massachusetts Institute of Technology, Cambridge, MA, USA.
¹⁴Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology, Cambridge, MA, USA.
¹⁵Department of Pathology, Stanford University, Stanford, CA, United States.
¹⁶Department of Statistics and Data Science, Yale University, New Haven, CT, United States.

⁰Ragon Institute of MGH, MIT, and Harvard, Cambridge, MA, USA.

Biorxiv : the Preprint Server for Biology +

|

May 27, 2024

View abstract on PubMed

Summary

Cell Spatio- and Neighborhood-informed Annotation and Patterning (CellSNAP) is a new self-supervised method for spatial-omics. It integrates molecular, positional, and image data for enhanced cell population discovery.

Area Of Science

Computational biology
Single-cell spatial-omics
Bioinformatics

Background

Accurate cell population identification is crucial for single-cell and spatial-omics studies.
Existing methods often fail to integrate molecular, positional, and imaging data from spatial-omics.
Reliance on reference datasets limits discovery to predefined cell types.

Approach

Introduced Cell Spatio- and Neighborhood-informed Annotation and Patterning (CellSNAP), a self-supervised computational method.
CellSNAP learns a fused representation vector for each cell by integrating molecular profiles, cellular neighborhood, and tissue image information.
The method operates at single-cell or finer resolution on spatial-omics data.

Key Points

CellSNAP enables *de novo* discovery of biologically relevant cell populations with fine granularity.
Demonstrated enhanced cell population discovery across spatial proteomic and transcriptomic modalities.
Successfully applied to diverse tissue types and disease settings.

Conclusions

CellSNAP effectively integrates multi-modal spatial-omics data for comprehensive cell identification.
The method overcomes limitations of existing approaches by leveraging spatial context.
Facilitates uncovering functionally stratified cell populations within their tissue microenvironment.

Keywords:

cellular neighborhoods clustering graph neural network multiplexed imaging spatial-omics