NEWS
SVG 1.0.0 (2026-02-01)
First Release
New Features
- Unified interface:
CalSVG() function provides a single entry point for all SVG detection methods
- Six SVG detection methods implemented with consistent output format:
CalSVG_MERINGUE(): Moran's I with binary spatial networkCalSVG_Seurat(): Moran's I with inverse distance weightsCalSVG_binSpect(): Binary spatial enrichment test (Giotto)CalSVG_SPARKX(): Non-parametric kernel-based testCalSVG_nnSVG(): Nearest-neighbor Gaussian processesCalSVG_MarkVario(): Mark variogram method
Spatial Network Construction
buildSpatialNetwork(): Build spatial neighborhood networks using Delaunay triangulation or KNNgetSpatialNeighbors_Delaunay(): Delaunay triangulation-based networkgetSpatialNeighbors_KNN(): K-nearest neighbors network
Statistical Utilities
moranI(): Calculate Moran's I statisticmoranI_test(): Hypothesis testing for spatial autocorrelationACAT_combine(): Aggregated Cauchy Association Test for p-value combinationbinarize_expression(): Multiple methods for expression binarization
Data Simulation
simulate_spatial_data(): Generate simulated spatial transcriptomics data with known SVGs- Support for multiple spatial patterns: gradient, hotspot, periodic, cluster
Performance
- C++ implementation via Rcpp/RcppArmadillo for computationally intensive operations
- Parallel processing support via
n_threads parameter
- Efficient memory usage for large-scale data
Documentation
- Comprehensive vignette with mathematical background
- Complete function documentation with examples
- Benchmark comparison between methods
Dependencies
- Core: Matrix, Rcpp, RcppArmadillo
- Optional: geometry, RANN, BRISC, CompQuadForm, BiocParallel, spatstat
Authors