ctest: Exact poisson test on the ratio between IP and input counts.

In exomePeak2: Bias Awared Peak Calling and Quantification for MeRIP-Seq

Description

Exact poisson test on the ratio between IP and input counts.

Usage

ctest(
  IP_count,
  input_count,
  IP_sizeFactor,
  input_sizeFactor,
  fold = 1,
  alpha = 0.05
)

Arguments

IP_count	a numeric vector for the counts of IP sample.
input_count	a numeric vector for the counts of input sample.
IP_sizeFactor	a numeric vector for the size factors of IP sample.
input_sizeFactor	a numeric vector for the size factors of input sample.
fold	the fold change under the null hypothesis; default = 1.
alpha	the alpha level used for determining the optimal independent filtering in filtered_R; default = 0.05.

Details

C-tests will be conducted between each entries of the IP and input vector.

The following statistical model is assumed on the data:

IP ~ Poisson(mean_IP*IP_sizeFactor)

input ~ Poisson(mean_input*input_sizeFactor)

The one-sided statistical test (C-test) is conducted with the following hypothesis pair:

H_0: mean(IP)/mean(input) <= fold

H_1: mean(IP)/mean(input) > fold

The exact p-values will be generated using binomial test, check poisson.test for more details.

The p-values are adjusted with "BH" method with an independent filtering slected by function filtered_R in package genefilter;

the filter statistics used is the normalized read abundence.

Value

A list include adjusted p values (with method fdr) and the corresponding log2 fold changes.

exomePeak2 documentation built on Nov. 8, 2020, 5:27 p.m.