test_independence: Fisher Test for 'hetset' Objects

In ZytoHMGU/hetset: Identification of Heterogeneous Subsets in Data

Description

Test for independence of subpopulations assigned from mixture model and classification of samples following supplemental information (e.g. sex, clinical groups)

Usage

test_independence(H, feature, S)

Arguments

H	hetset object with partitioning of samples (H$prt) as well as supplemental information stored as sample information in H$[name].
feature	name of the feature to be tested for independence
S	reference value for classification of samples according to feature which must be one of type logical, factor or double (e.g. S = TRUE, S = "MALE", S = 1.5)

Details

A partitioning of samples found by scan_hetset is tested for association with sample classes formed by external categories (e.g. sex, smoking, age > 50 years, etc.).

Fisher's exact test for independence is applied.

Value

Returns output of fisher.test function

Note

If scan_hetset was initiated by a given classification (e.g. sex), the test must not be executed for the partitioning found by the mixture model and the initial classification information. Since this would represent some kind of a tautology.

Author(s)

Daniel Samaga

Examples

n_A <- 100
n_B <- 50
A <- matrix(data = rnorm(n = n_A*15,mean = 1,sd = 1),ncol = n_A)
B <- matrix(data = rnorm(n = n_B*15,mean = 3,sd = 2),ncol = n_B)
Hds <- hetset(D = cbind(A,B))
Hds$"group" <- c(rep("A",n_A),rep("B",n_B))
Hds$"score" <- rnorm(ncol(Hds))
rm(A,B,n_A,n_B)
Hds <- scan_hetset(H = Hds,level = "univariate",min_size = 2,
    max_size = 3,rel_imp = 0.1,em_steps = 5)
test_independence(H = Hds,feature = "group","A")
test_independence(H = Hds,feature = "group","B")
test_independence(H = Hds,feature = "score",0)

ZytoHMGU/hetset documentation built on June 6, 2019, 2:16 p.m.