fdr.test {bayesclust} | R Documentation |
Allows the experimenter to assess the significance of clusters in multiple datasets whilst controlling for the False Discovery Rate (FDR).
fdr.test(namelist, q=0.05)
namelist |
The initial argument should be a character vector of names
of objects of class ``emp2pval''. Objects of this class are returned by the function
emp2pval . |
q |
q should be a value between 0 and 1. It is the maximum FDR
that the experimenter is willing to tolerate. |
This function implements the FDR controlling procedure described in Benjamini and Hochberg (1995). This routine will look for the largest i such that P_{(i)} <= (i/m) * q . Please refer to the original paper for the notation and details.
The output simply lists the datasets for which significant clusters were found.
Gopal, V.
Benjamini, Y. and Hochberg, Y. (1995) Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. JRRS (B) Vol. 57, No. 1
emp2pval
for details on objects of class emp2pval
.
# Generate random 2-variate data Y <- matrix(rnorm(24), nrow=12) Z <- matrix(rnorm(24), nrow=12) # Search for optimal partitioning of data into 2 clusters test1 <- cluster.test(Y, p=2) test2 <- cluster.test(Z, p=2) # Generate corresponding null density object. null1 <- nulldensity(nsim=100, n=12, p=2, k=2) # Convert EPP to p-value test1.pval <- emp2pval(test1, null1) test2.pval <- emp2pval(test2, null1) # Test for significance, controlling for FDR fdr.test(c("test1.pval", "test2.pval"), q=0.3)