mlog10p: Negative log10 P values
In tobyjohnson/gtx: Genetics ToolboX

Description Usage Arguments Details Value Author(s) Examples

Compute minus log10 P values from effect sizes and standard errors.

1	mlog10p(beta, se)

`beta`	a vector of effect sizes
`se`	a vector of standard errors

The results are the same as -log10(pnorm(-abs(beta)/se)*2), but are calculated to machine precision even when the P values are very small (such that the intermediate step of calculating the P values would give zero to machine precision).

The vector of minus log10 P values.

Toby Johnson Toby.x.Johnson@gsk.com

beta <- c(1:4, (1:10)*5) * rep(c(-1, 1), 7)
se <- rep(1, length(beta))
pval1 <- pnorm(-abs(beta)/se)*2
pval2 <- pchisq(beta^2/se^2, df = 1, lower.tail = FALSE)
cbind(z = beta/se, pval = pval1, -log10(pval1), -log10(pval2), mlog10p(beta, se))