BioGSP-package: BioGSP: Biological Graph Signal Processing for Spatial Data...
In BioGSP: Biological Graph Signal Processing for Spatial Data Analysis

BioGSP-package

R Documentation

BioGSP: Biological Graph Signal Processing for Spatial Data Analysis

Description

The BioGSP package provides a comprehensive implementation of Graph Signal Processing (GSP) methods including Spectral Graph Wavelet Transform (SGWT) for analyzing spatial patterns in biological data. This implementation is based on Hammond, Vandergheynst, and Gribonval (2011) "Wavelets on Graphs via Spectral Graph Theory".

Details

The package enables multi-scale analysis of spatial signals by:

Building graphs from spatial coordinates using k-nearest neighbors
Computing graph Laplacian eigendecomposition for spectral analysis
Designing wavelets in the spectral domain using various kernel functions
Decomposing signals into scaling and wavelet components at multiple scales
Providing reconstruction capabilities with error analysis
Offering comprehensive visualization and analysis tools

Main Functions

initSGWT: Initialize SGWT object with data and parameters
runSpecGraph: Build graph and compute eigendecomposition
runSGWT: Perform forward and inverse SGWT transforms
runSGCC: Calculate weighted similarity between signals
sgwt_forward: Forward SGWT transform
sgwt_inverse: Inverse SGWT transform
sgwt_energy_analysis: Energy distribution analysis
plot_sgwt_decomposition: Visualization of SGWT components
demo_sgwt: Demonstration with synthetic data

Applications

The BioGSP package is particularly useful for:

Spatial biology: Analyzing cell distribution patterns in tissue imaging (CODEX, Visium, etc.)
Single-cell genomics: Spatial transcriptomics and proteomics analysis
Neuroscience: Brain connectivity and signal analysis
Pathology: Tumor microenvironment and tissue architecture analysis
Developmental biology: Spatial pattern formation and cell fate mapping
Immunology: Immune cell spatial organization and interactions

Author(s)

BioGSP Development Team

References

Hammond, D. K., Vandergheynst, P., & Gribonval, R. (2011). Wavelets on graphs via spectral graph theory. Applied and Computational Harmonic Analysis, 30(2), 129-150.

Examples


# Load the package
library(BioGSP)

# Run a quick demo
demo_result <- demo_sgwt()

# Generate synthetic data
set.seed(123)
n <- 100
data <- data.frame(
  x = runif(n, 0, 10),
  y = runif(n, 0, 10),
  signal = sin(runif(n, 0, 2*pi))
)

# New workflow: Initialize -> Build Graph -> Run SGWT
SG <- initSGWT(data, signals = "signal", J = 4, kernel_type = "heat")
SG <- runSpecGraph(SG, k = 8)
SG <- runSGWT(SG)

# Analyze results
energy_analysis <- sgwt_energy_analysis(SG)
print(energy_analysis)

BioGSP documentation built on Feb. 2, 2026, 5:06 p.m.