Minimums in ascertain_pedigrees seem to be swapped or overlapped

[WValenti]

In ascertain peds. If I ask for min 3 affecteds and min 4 individuals in ped, I see only peds with at least 4 affecteds. If I ask for min 3 affecteds and min 3 individuals, I do see peds with exactly 3 affecteds. So the "min individuals" constraint appears to be doing the same thing as the "min affecteds" constraint, at least in this case.

[Viqsi]

Fixed in MathematicalMedicine/diverRPC@6d15720

[Viqsi]

Well, crud. Referring to both of those parameters as "minimums" screwed up understanding and thus screwed up the fix.

There are THREE parameters involved:

• A number of individuals for pedigree size, which is supposed to be a minimum.
• A number of individuals for variable constraint, which can be BUT IS NOT NECESSARILY a minimum.
• what type of numeric constraint the variable constraint uses - minimum, maximum, or exact.

We accidentally swapped not just which count is which, but also which count should be getting the user-specified constraint type. And when I "fixed" it, I swapped only the count parameters. D'oh!

Figuring this out obliged me to peek yet again at the original design doc in #134 to make sure it's supposed to be "min individuals" and "min/max/exact affected". Here it is as described by Veronica:

Include all pedigrees with [at least/at most/exactly] [2, 3,...] individuals with constraint = [show drop down list of values for selected constraining variable] and at least [2, 3,…] total number of individuals.

(There's also a specification in there for non-categorical/dichotomous variables, but that is unimplemented; we're Categorical/Dichotomous Only for now, still.)