selectRho: Choosing a threshold based on the...

In GOGANPA: GO-Functional-Network-based Gene-Set-Analysis

Description

Determine the threshold parameter which will result in a network with optimal scale-free fitness.

Usage

selectRho(simMat, rhovec = NULL)

Arguments

simMat	The GO-similairty matrix. Missing and negative entries are not allowed. The gene names should be assigned to the row and column names.
rhovec	a vector of candidate thresholds, or if NULL, a set of thresholds chosen according to the range of the similarity matrix.

Details

The scale-free fitness measure is based on linear-regression-based R-squared goodness-of-fit measure.

Value

A list, with elements:

criterion	a summary table of the candidate thresholds' resulting fits.
bestrho	The candidate threshold with the highest R-squared.

Note

Note that certain GO-similarity measures are unbounded (e.g. the Resnik similarity). This code will not normalize the similarity matrix, and rhovec, if supplied, should be chosen according to the range of the GO-similarity values inside simMat.

Author(s)

Billy Chang

References

Chang, B., Kustra, R. and Tian, WD (2012) Functional-Network-based Gene Set Analysis using Gene Ontology. Submitted.

Zhang, B. and Horvath, S. (2005) A General Framework for Weighted Gene Co-Expression Network Analysis. Statistical Applications in Genetics and Molecular Biology. 4:1:A17.

See Also

getGNET

Examples

#Not to Run
data("simMatSmall",package="GOGANPA")
fit <- selectRho(simMatSmall)
plot(fit$criterion[,1],fit$criterion[,2])
abline(v=fit$bestrho,col=2)

GOGANPA documentation built on May 2, 2019, 6:04 a.m.