This is unnecessarily conservative in the special case of direction="any", lfc >> 0 and block=NULL. But I don't want to handle that special case, and I can't figure out how to generalize the less-conservative approach to survive Stouffer's method.
In some respects, the conservativeness of the special case doesn't really matter, given how low the p-values are anyway.
From discussion in #86.
This is unnecessarily conservative in the special case of
direction="any",lfc >> 0andblock=NULL. But I don't want to handle that special case, and I can't figure out how to generalize the less-conservative approach to survive Stouffer's method.In some respects, the conservativeness of the special case doesn't really matter, given how low the p-values are anyway.