compute_differential_evidence: Compute a null distribution for function values on a region...

In lcrawlab/SINATRA: Sub Image Analysis using Topological Summary Statistics (SINATRA)

Compute a null distribution for function values on a region of a mesh

Description

Input a region on the mesh, and the function finds N other connected regions of the same size, uniformly sampled from the mesh. The goal is to compare the function values on the input region to those on the N other regions in order to generate the analog of p values.

Usage

compute_differential_evidence(
  complex,
  mesh_fcn,
  vertex,
  region_size = 100,
  num_test_regions = 100,
  method = "knn"
)

Arguments

mesh_fcn	(1 x M vector) the values of the function on the vertices of the mesh.
vertex	(int) the index of the desired vertex to analyze on the mesh
region_size	(int) the number of vertices to be in the region. We choose the region via knn. The assumption is that the density if vertices in the mesh is uniform, or at least approximately.
num_test_regions	(int) the number of regions to sample from
method	(string) the method to use when creating null regions. We currently support 'area' and 'knn'.
mesh	mesh3d object.

Value

sum of function values for the input region, and a vector of length N of the sum of function values for the random test regions

lcrawlab/SINATRA documentation built on Sept. 13, 2023, 2 p.m.