metaSeq (version 1.12.0)

other.oneside.pvalues: Optional function for non-NOISeq method

Description

Optional function for non-NOISeq method users. P-values or probability in one-sided test in positive and negative differentiation is integrated and converted as a input matrix for Fisher.test or Stouffer.test. Weight in each study can also be introduced.

Usage

other.oneside.pvalues(Upper, Lower, weight = NULL)

Arguments

Upper

A matrix which means p-values or probability in one-sided test (positive differentiation). Its rows mean "gene" and its columns mean "study" that test was conducted.

Lower

A matrix which means p-values or probability in one-sided test (negative differentiation). Its rows mean "gene" and its columns mean "study" that test was conducted.

weight

A vector which means weight in each study. Its length must be equal to the number of column in Upper and Lower.

Examples

## Assume these are one-sided p-value generated by non-NOISeq method (e.g., cufflinks)
upper <- matrix(runif(300), ncol=3, nrow=100)
lower <- 1 - upper
rownames(upper) <- paste0("Gene", 1:100)
rownames(lower) <- paste0("Gene", 1:100)
weight <- c(3,6,8)

# other.oneside.pvalues function return a matrix which can input Fisher.test or Stouffer.test
result <- other.oneside.pvalues(upper, lower, weight)

# Fisher's method (without weighting)
F <- Fisher.test(result)
str(F)

# Stouffer's method (with weighting by sample-size)
S <- Stouffer.test(result)
str(S)