An improved hierarchical variational autoencoder for cell-cell communication estimation using single-cell RNA-seq...

Area of Science:

• Computational biology
• Genomics
• Cancer research

Background:

• Cell-cell communication (CCC) in the tumor microenvironment is crucial for understanding cancer progression and drug tolerance.
• Single-cell RNA sequencing (scRNA-Seq) offers vast data for predicting cellular interactions.
• Existing methods for inferring CCC rely on limited prior knowledge, leading to inaccuracies.

Purpose of the Study:

• To develop an improved model for estimating CCC by leveraging comprehensive scRNA-Seq data.
• To overcome limitations of existing methods that depend on incomplete molecular interaction databases.
• To provide a more accurate and automated approach for analyzing cellular communication.

Main Methods:

• Proposed an improved hierarchical variational autoencoder (HiVAE) model.
• Utilized HiVAE to learn cell representations from both known ligand-receptor genes and all genes in scRNA-Seq data.
• Employed transfer entropy to quantify information flow and estimate directed communication relationships between cells.

Main Results:

• The HiVAE model effectively learned meaningful cell representations.
• Transfer entropy accurately estimated communication scores between cell types.
• Demonstrated the model's efficacy on human skin disease and melanoma datasets.

Conclusions:

• The HiVAE model offers a powerful approach for automated CCC estimation from scRNA-Seq data.
• This method enhances the understanding of cellular interactions in complex biological systems.
• The findings have implications for cancer research and therapeutic development.