Skip to contents

SpaGER

📖 Documentation: https://zaoqu-liu.github.io/SpaGER/

Spatial Gene Enhancement via Domain Adaptation

SpaGER (Spatial Gene Expression in R) provides a high-performance R implementation of the SpaGE algorithm for predicting genome-wide expression profiles in spatial transcriptomics data through integration with single-cell RNA sequencing (scRNA-seq) reference datasets.

Spatial transcriptomics technologies, while providing invaluable spatial context, often measure only a limited panel of genes. SpaGER addresses this fundamental limitation by leveraging the comprehensive gene coverage of scRNA-seq data to impute unmeasured gene expression in spatial data, enabling more complete transcriptomic characterization of tissue architecture.

Theoretical Framework

SpaGER employs a domain adaptation approach based on Principal Vectors (PVs) to bridge the gap between spatial and single-cell transcriptomic domains:

Mathematical Foundation

The algorithm operates through the following steps:

  1. Z-score Standardization: Both datasets are standardized using population standard deviation to ensure comparability

  2. Principal Vector Computation: Identifies the optimal linear subspace that maximizes alignment between domains through SVD-based decomposition

  3. Cosine Similarity Selection: Principal vectors are retained based on their cosine similarity, representing the alignment quality between domains

  4. Weighted k-NN Imputation: Expression values are predicted using distance-weighted averaging of k nearest neighbors in the aligned subspace

Key Features

  • High-Performance Computing: C++ acceleration via Rcpp/RcppArmadillo for computationally intensive operations
  • Numerical Accuracy: Mathematically equivalent results to the original Python implementation (tolerance < 1e-6)
  • Seurat Integration: Native support for Seurat v4 and v5 objects with automatic version detection
  • Cross-Platform Compatibility: Validated on Windows, macOS (Intel/ARM), and Linux
  • Flexible Dimensionality Reduction: Support for PCA, ICA, Factor Analysis, NMF, and PLS methods
  • Comprehensive Validation: 60+ unit tests ensuring scientific accuracy

Installation 

# Install from R-Universe (recommended)
install.packages("SpaGER", repos = "https://zaoqu-liu.r-universe.dev")

Alternative: GitHub 

# Using remotes
remotes::install_github("Zaoqu-Liu/SpaGER")

# Or using devtools
devtools::install_github("Zaoqu-Liu/SpaGER")

Optional Dependencies

For extended functionality:

install.packages(c("Seurat", "fastICA", "NMF", "pls", "psych"))

Quick Start 

library(SpaGER)

# Predict expression of unmeasured genes
predicted <- SpaGE(
  spatial_data = spatial_expr,    # Spatial transcriptomics data
  rna_data = scrna_expr,          # scRNA-seq reference
  n_pv = 30,                      # Number of principal vectors
  n_neighbors = 50                # k for KNN imputation
)

# With Seurat objects
library(Seurat)
spatial_obj <- SpaGE.Seurat(
  spatial_seurat = visium_data,
  rna_seurat = scrna_ref,
  n_pv = 30
)

See the documentation for comprehensive tutorials and examples.

Citation

If you use SpaGER in your research, please cite:

Original Algorithm

Abdelaal T, Mourragui S, Mahfouz A, Reinders MJT. (2020). SpaGE: Spatial Gene Enhancement using scRNA-seq. Nucleic Acids Research, 48(18):e107. DOI: 10.1093/nar/gkaa740

R Implementation

Liu Z. (2026). SpaGER: Spatial Gene Expression Prediction using scRNA-seq in R. GitHub: https://github.com/Zaoqu-Liu/SpaGER

License

This project is licensed under the MIT License.

Contact