An Intrinsic-hoc Framework for Heterogeneous Cellular Senescence Elucidation Using Deep Graph Representation Learning...

Area of Science:

• Genomics
• Computational Biology
• Aging Research

Background:

• Cellular senescence plays a key role in aging and age-related diseases.
• Identifying senescent cells and their gene expression profiles in complex tissues is challenging.
• Understanding senescence heterogeneity is crucial for therapeutic development.

Purpose of the Study:

• To introduce DeepSAS, a deep graph representation learning framework.
• To elucidate senescent cell heterogeneity and associated genes from single-cell RNA-seq data.
• To validate the framework's performance on real-world datasets.

Main Methods:

• Developed DeepSAS, an intrinsic-hoc framework utilizing deep graph representation learning.
• Applied DeepSAS to single-cell RNA-seq data from healthy eye and idiopathic pulmonary fibrosis datasets.
• Validated findings using Xenium spatial transcriptomics.

Main Results:

• DeepSAS effectively identified robust and biologically grounded senotypes.
• The framework demonstrated superior performance compared to existing methods in benchmarking tests.
• Revealed heterogeneity within senescent cell populations.

Conclusions:

• DeepSAS provides a powerful tool for characterizing senescent cells and their associated genes.
• The framework enhances our understanding of cellular senescence in aging and disease.
• Spatial transcriptomics validation supports the biological relevance of identified senotypes.