Learn R Programming

gap (version 1.6)

metap: Meta-analysis of p values

Description

Meta-analysis of p values

Usage

metap(data, N, verbose = "Y", prefixp = "p", prefixn = "n")

Value

  • x2 Fisher's chi-squared statistics.

  • p P values from Fisher's method according to chi-squared distribution with 2*N degree(s) of freedom.

  • z Combined z value.

  • p1 One-sided p value.

  • p2 Two-sided p value.

Arguments

data

data frame.

N

Number of studies.

verbose

Control of detailed output.

prefixp

Prefix of p value, with default value "p".

prefixn

Preifx of sample size, with default value "n".

Author

Jing Hua Zhao

Details

This function is the method of meta-analysis used in the Genetic Investigation of ANThropometric Traits (GIANT) consortium, which is based on normal approximation of p values and weighted by sample sizes from individual studies.

See Also

metareg

Examples

Run this code
if (FALSE) {
s <- data.frame(p1=0.1^rep(8:2,each=7,times=1),n1=rep(32000,49),
                p2=0.1^rep(8:2,each=1,times=7),n2=rep(8000,49))
cbind(s,metap(s,2))

# Speliotes, Elizabeth K., M.D. [ESPELIOTES@PARTNERS.ORG]
# 22-2-2008 MRC-Epid JHZ

np <- 7
p <- 0.1^((np+1):2)
z <- qnorm(1-p/2)
n <- c(32000,8000)
n1 <- n[1]

s1 <- s2 <- vector("numeric")

for (i in 1:np)
{
   a <- z[i]
   for (j in 1:np)
   {
       b <- z[j]
       metaz1 <- (sqrt(n1)*a+sqrt(n[1])*b)/sqrt(n1+n[1])
       metap1 <- pnorm(-abs(metaz1))
       metaz2 <- (sqrt(n1)*a+sqrt(n[2])*b)/sqrt(n1+n[2])
       metap2 <- pnorm(-abs(metaz2))
       k <- (i-1)*np+j
       cat(k,"\t",p[i],"\t",p[j],"\t",metap1,metaz1,"\t",metap2,metaz2,"\n")
       s1[k] <- metap1
       s2[k] <- metap2
  }
}

q <- -log10(sort(p,decreasing=TRUE))
t1 <- matrix(-log10(sort(s1,decreasing=TRUE)),np,np)
t2 <- matrix(-log10(sort(s2,decreasing=TRUE)),np,np)

par(mfrow=c(1,2),bg="white",mar=c(4.2,3.8,0.2,0.2))
persp(q,q,t1)
persp(q,q,t2)
}

Run the code above in your browser using DataLab