mcc p-value is not correct for negative estimate

[spocks]

For negative values of mcc estimate, function is returning wrong p-values

x <- factor(c("A", "B", "A", "B", "A", "A"))
CoreGx::mcc(x,x)

#$estimate
#[1] 1
#$p.value
#[1] 0.000999001

y <- factor(x, levels = c("B","A"))
CoreGx::mcc(x, y)

#$estimate
#[1] -1
#$p.value
#[1] 0.941

[p-smirnov]

No one said it was a two sided p value :smile:

Will fix with the standard alternative parameter

[bhaibeka]

Should we make it two-sided by default?

[spocks]

yes should be two-sided by default just like Pearson etc.

[p-smirnov]

22a4502d65f9bc5976a877ed0f0edc3b5dd1ab88 is not two sided (missing a factor of 2)

[p-smirnov]

09e5658 adds alternative par, defaults to two sided.

good catch on the >= @spocks