A unified analysis of atlas single-cell data

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Atlas-scale single-cell data generation enables cross-tissue and cross-modality analysis.
• Current methods focus on cells as reference units, limiting gene-centric analyses and biological insights.
• Existing approaches may obscure genuine biological variation present in large datasets.

Purpose of the Study:

• To introduce GIANT, an integration method for atlas-scale gene analysis across diverse cell types and human tissues.
• To overcome limitations of cell-centric approaches in large-scale genomic data integration.
• To facilitate comprehensive gene-based analysis and discovery.

Main Methods:

• GIANT converts datasets into gene graphs for analysis.
• The method employs recursive gene embedding without requiring additional alignment.
• It processes atlas-scale single-cell data from multiple tissues and modalities.

Main Results:

• GIANT successfully generated unified gene-embedding spaces across human tissues and data modalities.
• Application to two large atlas datasets demonstrated the method's efficacy.
• The approach facilitated the discovery of diverse gene functions and regulatory mechanisms.

Conclusions:

• GIANT provides a powerful gene-centric approach for integrating atlas-scale data.
• The method enhances the discovery of gene functions and regulatory networks.
• GIANT is valuable for analyzing complex biological variation across tissues and modalities.