Correspondence between constraint tree and alignment

[rvosa]

This is not so much an issue but an elaboration that perhaps should end up in various reports and documentation files. This is about how the tips in a constraint tree must logically correspond with the sequences in an alignment. When errors occur surrounding this topic, then this might be a reference in trying to resolve the issue. Here are some possible patterns in which tree and alignment relate to each other:

1. There is no constraint tree at all. This is actually the most common way in which raxml is used. You could rephrase this situation as one where there is a constraint tree but it has zero tips. That may sound trivial but it's worth keeping in mind that the tool normally places sequences in a tree based only on the signal that's in the DNA so there's no need for any tips to be in a constraint.
2. There's a constraint tree with some taxa from the alignment. In this case the constraint tree has a subset of the tips that are in the alignment. This is fine because it just means that there is a bit more signal besides the DNA, for example to make it so that a few tips are forced to be next to each other. This is basically what we're going for.
3. The constraint tree has tips that are not in the alignment. This is impossible so it produces errors. The tree searcher is now being asked to place tips in a tree but it has no DNA data to do so. Tips in the constraint tree absent from the alignment must be pruned.
4. The constraint tree or the alignment has multiple entries with the same name. This is also impossible. The solution here is to make the constraint tree have a polytomy (e.g. ((A1,A2,A3,A4),...)) and make the alignment also have A1, A2, A3 and A4 (instead of multiple, identical As).

I think that's all the possible arrangements :)