SPRITE: improving spatial gene expression imputation with gene and cell networks

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Spatially resolved single-cell transcriptomics offers insights into in situ gene expression and tissue organization.
• Current spatial transcriptomics technologies are limited to measuring a small number of genes.
• Existing algorithms for gene expression imputation do not fully utilize spatial and gene relational information.

Purpose of the Study:

• To introduce SPRITE (Spatial Propagation and Reinforcement of Imputed Transcript Expression), a meta-algorithm designed to improve spatial gene expression predictions.
• To leverage rich spatial and gene relational information in spatial transcriptomics data.
• To enhance the accuracy of gene expression imputation for a broader gene panel.

Main Methods:

• SPRITE processes predictions from existing imputation methods.
• It propagates information across gene correlation networks.
• It utilizes spatial neighborhood graphs to refine predictions.

Main Results:

• SPRITE significantly improves spatial gene expression predictions across diverse spatial transcriptomics datasets.
• Predicted spatial gene expression enhances cell clustering, visualization, and classification.
• SPRITE integration leads to more robust inferences in spatial transcriptomics data analysis.

Conclusions:

• SPRITE is an effective meta-algorithm for enhancing spatial gene expression imputation.
• The method improves downstream analyses in spatial transcriptomics.
• SPRITE offers a valuable tool for researchers in the field.