Deconvolution and inference of spatial communication through optimization algorithm for spatial transcriptomics.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Spatial transcriptomics provides gene expression data at specific tissue locations.
• Current deconvolution methods often overlook spatial information and cell-cell communication.
• This limits accurate cell-type inference and understanding of tissue organization.

Purpose of the Study:

• To develop a novel deconvolution algorithm, NODE, that incorporates spatial information and intercellular communication.
• To improve cell-type inference accuracy in spatial transcriptomics data.
• To enable simultaneous quantification of intercellular communications within tissues.

Main Methods:

• Proposed NODE (non-negative least squares-based and optimization search-based deconvolution) algorithm.
• Combined cell-type information from single-cell RNA sequencing (scRNA-seq) with intercellular communication data.
• Utilized optimization methods to infer deconvolution results and spatial communications.

Main Results:

• NODE accurately deconvolves spatial transcriptomics data, outperforming existing methods.
• Successfully inferred and validated spatial communications in human heart development.
• Demonstrated reduced overfitting probability compared to traditional algorithms.

Conclusions:

• NODE is a robust deconvolution method for spatial transcriptomics, enhancing cell-type and communication analysis.
• The integration of spatial context and intercellular communication provides deeper biological insights.
• NODE facilitates a more comprehensive understanding of tissue architecture and function.