checkKband: Check K-band limited property of signals

In BioGSP: Biological Graph Signal Processing for Spatial Data Analysis

Check K-band limited property of signals

Description

Analyze whether signals are k-band limited by comparing low-frequency and high-frequency Fourier coefficients using eigendecomposition and statistical testing. Builds graph and computes Laplacian directly from SGWT data.

Usage

checkKband(
  SG,
  signals = NULL,
  alpha = 0.05,
  verbose = TRUE,
  k = 25,
  laplacian_type = "normalized"
)

Arguments

SG	SGWT object with Data slot (from initSGWT)
signals	Character vector of signal names to analyze. If NULL, uses all signals from SG$Data$signals
alpha	Significance level for Wilcoxon test (default: 0.05)
verbose	Logical; if TRUE, print progress messages (default: TRUE)
k	Number of nearest neighbors for graph construction (default: 25)
laplacian_type	Type of Laplacian ("unnormalized", "normalized", or "randomwalk") (default: "normalized")

Value

List containing:

is_kband_limited

Logical; TRUE if all signals are k-band limited

knee_point_low

Integer; knee point index for low-frequency eigenvalues

knee_point_high

Integer; knee point index for high-frequency eigenvalues

signal_results

List with per-signal test results including p-values and Fourier coefficients

Examples

# Create example data
data <- data.frame(x = runif(100), y = runif(100),
                  signal1 = rnorm(100), signal2 = rnorm(100))

# Initialize SGWT object (no need to run runSpecGraph)
SG <- initSGWT(data, signals = c("signal1", "signal2"))

# Check k-band limited property
result <- checkKband(SG, signals = c("signal1", "signal2"), k = 30)
if (result$is_kband_limited) {
  cat("All signals are k-band limited")
}

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