[BUG]: inferred p-value calculation is missing a factor of 2

[j-brody]

Please make sure that this is a bug! If you have questions about how to use TwoSampleMR please use the Discussions function instead.

Describe the bug (required)

For 'inferred', the calculation for dat$pval.outcome is missing a factor of 2 in the p-value formula.

Describe the current behaviour you observe (required)

When the outcome p-values are missing, but beta.outcome and se.outcome are present, the following code is used to infer p-values:

if (any(is.na(dat$pval.outcome))) {
if ("beta.outcome" %in% names(dat) && "se.outcome" %in% names(dat)) {
index <- is.na(dat$pval.outcome)
dat$pval.outcome[index] <- stats::pnorm(abs(dat$beta.outcome[index])/dat$se.outcome[index],
lower.tail = FALSE)
dat$pval_origin.outcome[index] <- "inferred"
}
}
However, this assignment does not multiply the result by 2 (for a two-sided test), resulting in half the correct p-value.

Describe the behaviour you expect (required)

dat$pval.outcome[index] <- 2 * stats::pnorm(abs(dat$beta.outcome[index])/dat$se.outcome[index], lower.tail = FALSE)

R code to reproduce the issue (required)

dat <- data.frame(
beta.outcome = c(1.1, 0.5),
se.outcome = c(0.6, 0.3),
pval.outcome = c(NA, NA)
)

if (any(is.na(dat$pval.outcome))) {
if ("beta.outcome" %in% names(dat) && "se.outcome" %in% names(dat)) {
index <- is.na(dat$pval.outcome)
dat$pval.outcome[index] <- stats::pnorm(abs(dat$beta.outcome[index])/dat$se.outcome[index], lower.tail = FALSE)
dat$pval_origin.outcome[index] <- "inferred"
}
}
print(dat)

Compare with:

dat$pval.outcome <- 2 * stats::pnorm(abs(dat$beta.outcome)/dat$se.outcome, lower.tail = FALSE)
print(dat)

Please provide a minimal code snippet that will reproduce this issue

Contribute a solution (optional)

Please submit a pull request and/or briefly describe your proposed solution

System information

Please provide details of your operating system and R version

Additional context

Add any other context about the problem here