SpaGE

SpaGE predicts whole-transcriptome spatial gene expression by integrating single-cell RNA sequencing (scRNA-seq) data with spatial transcriptomics to infer unmeasured gene expression across spatially resolved cells.

Key Features:

• Integration of Datasets: Integrates spatial transcriptomics data, which provides spatial context but typically measures tens to hundreds of transcripts, with scRNA-seq data that provides whole-transcriptome expression without spatial coordinates.
• Whole-Transcriptome Prediction: Predicts full-transcriptome expression profiles within spatially resolved cells by leveraging scRNA-seq as a reference.
• Principal Vectors (PVs) Utilization: Employs principal vectors (PVs) for the integration process with the number of PVs specified by the user.
• Optional Gene Set Specification: Allows optional specification of a set of unmeasured genes in the spatial data for which predictions are sought from the scRNA-seq data.
• Scalability and Performance: Demonstrated superior performance on five dataset-pairs compared to previously published methods and scales to large datasets.
• Validation of Predictions: Predicted spatial gene patterns have been validated using in situ hybridization data from the Allen Mouse Brain Atlas.

Scientific Applications:

• Cellular heterogeneity mapping: Mapping cellular heterogeneity in complex tissues by imputing unmeasured genes into spatial context using scRNA-seq references.
• Brain tissue analysis: Resolving spatial gene expression patterns in the brain with validation against in situ hybridization data from the Allen Mouse Brain Atlas.
• Development and disease studies: Investigating developmental processes, disease mechanisms, and tissue architecture by providing spatially resolved, full-transcriptome profiles.

Methodology:

Integrates spatial transcriptomics and scRNA-seq data and predicts full transcriptome expression in spatial profiles using principal vectors (PVs) with a user-specified number and an optional specified set of unmeasured genes.