Closed hyanwong closed 1 year ago
The verbose description is something like "robust to HMM coordinate reflection", or something. What it's doing is saying that we have to chose sequence-identical parents when we have mirrored and unmirrored coordinates - given that we have the same number of parents and the mutations are the same, those parents must be identical in state. We don't insist on the parent nodes being the same because this is overly restrictive (when we have identical parents to choose from, the HMM will choose arbitrarily).
Does that help?
I like "robust" (much preferable to "simple"), but it may give the wrong impression that these are "good" (or "True") recombinants. I think @szhan was thinking about good terminology to use.
I also prefer "robust". I have been calling them "robustly identified" in the MM section.
I worry that "robust" will be interpreted as "good" or "confident". We could say something like "HMM orientation consistent"? We probably don't say it that much, so something precise might be preferable?
Let's just go with "HMM-consistent" in the preprint for now.
I'll use this term in the code also when implementing #108
I think we have fixed on "HMM-consistent" now
We currently define a "consistent" recombinant in the following way:
The word "consistent" is confusing here. We should come up with a better term.