Multtest NPM | npm.io

multtest

adjustments of p-values for multiple comparisons

Installation

npm install multtest

Usage:

Require as follows:

var multtest = require('multtest');

multtest exports the following functions:

`.bonferroni(pvalues,[numHypotheses])`

Given an input array of pvalues, pvalues, this function calculates the Bonferroni correction by multiplying each p-value by m, the number of tested hypotheses. This is by default equal to the length of the pvalues array, but can be optionally supplied via the numHypotheses parameter.

`.fdr(pvalues,[numHypotheses])`

Given an input array of pvalues, pvalues, the .fdr function calculates the false-discovery-rate adjusted p-values.

`.bY(pvalues,[numHypotheses])`

Given an input array of pvalues, pvalues, the .bY function calculates adjusted p-values according to the method by Benjamini & Yekutieli.

`.adjustSignificanceLevel(pvalues, alpha)`

This function can be used in the construction of FDR adjusted confidence intervals. It has two parameters: pvalues is an array of p-values, alpha is the significance level we wish to control the FDR at. The function returns an adjusted signficance level alpha_fdr which has to be used as the nominal significance level when constructing confidence intervals. It is calculated via the formula

alpha_fdr = ( (k + 1) / m ) * alpha,

where m is the total number of hypotheses and k is the number of rejected hypotheses.