Run multi version of skrCCA to detect subsequent components (Single Slide) Uses flat kernel structure for consistent data access

Run multi version of skrCCA to detect subsequent components (Single Slide) Uses flat kernel structure for consistent data access

Usage

optimize_bilinear_n(
  X_list,
  flat_kernels,
  sigma,
  w_list,
  cellTypesOfInterest,
  nCC = 2,
  max_iter = 1000,
  tol = 1e-05,
  step_size = 1,
  sdev2_list = NULL
)

Arguments

X_list

Named list of data matrices (subsetted)

flat_kernels

Flat list of kernel matrices

sigma

Sigma value (numeric)

w_list

A named list of weights (subsetted, matrices with previous components as columns)

cellTypesOfInterest

A vector specifying cell type names present in the input lists

nCC

Total number of canonical vectors desired (must be >= 2)

max_iter

Maximum number of iterations for helper function

tol

Tolerance of accuracy for helper function

step_size

Step size for damped power iteration (default 1)

sdev2_list

Optional named list of squared standard deviations per cell type for weighted normalization. Default NULL.

Value

A named list of weights (matrices with components 1 to nCC as columns)