option for qnode / knode mapping

[cbizon]

Can a single knode map to two qnodes in the same answer? In robokop, the answer was yes, and this turned out to be useful sometimes, when you didn't want to force two nodes to be the same, but you could allow it. For instance, say you have two chemicals you want to relate to a common unknown disease via unknown genes:
(chem1)--(genea)--(disease)--(geneb)--(chem2).
If genea and geneb end up being the same, then that's actually a better, more parsimonious answer.

But sometimes, it's annoying. Say for instance that you want to find a drug for diseaseX by looking for a drug that treats a phenotypically similar drug:
(diseaseX)--(set of phenotypes)--(disease)--(drug)
If you don't control the k/q mapping, disease=diseaseX is a valid answer, but the results are not helpful.

So I think that this may need to be an option. I could imagine worse cases where you want the option set one way for one set of nodes and a different way for others. i.e. the ability to add arbitrary a<>b cypher clauses.

[patrickkwang]

Quick question: Do we care to have the same options for edges?

[cbizon]

I can't think of a case... I've certainly written cypher queries that distinguish nodes, but never worried about edges that I can recall.