Add omit_missing to Tree.nodes()?

[hyanwong]

At the moment, the Tree.nodes() method lists all the nodes in a tree, even if the node is isolated (i.e. has missing data). It would be useful for me to somehow iterate through the nodes without counting the missing ones. A parameter to the Tree.nodes() iterator would work, or perhaps an easy way to check if a node is isolated? The only way I can think to do this now is to check Tree.num_children(node) == 0 and Tree.parent(node) == tskit.NULL - is that right?

[jeromekelleher]

I guess a first step is to add a method Tree.is_isolated(u), which does the check of no parent and no children. I'm not sure we want to add a parameter to the nodes iterator, since it's pretty easy to filter the nodes out afterwards (given this method) and there's no performance advantage.

[hyanwong]

Fine. I guess it is useful to have the Tree.is_isolated(u) method, thought, rather than the more unwieldy version I typed above. And this should be done in pure Python?

[hyanwong]

Oh, and it might be thought a bit odd that isolated nodes ("not in any tree") are returned by Tree.nodes(), so perhaps it's worth a comment somewhere in the docs, either in the Tree.nodes() docstring or in the section on missing data?

[jeromekelleher]

Oh, and it might be thought a bit odd that isolated nodes ("not in any tree") are returned by Tree.nodes(), so perhaps it's worth a comment somewhere in the docs, either in the Tree.nodes() docstring or in the section on missing data?

They are in the tree though --- they are roots. These are isolated samples, not just any old nodes. So, we need to be careful with the terminology. I don't think we need to over elaborate on this, it'll likely just confuse people.

And this should be done in pure Python?

Sure, no point in complicating this.

[hyanwong]

They are in the tree though --- they are roots. These are isolated samples, not just any old nodes. So, we need to be careful with the terminology. I don't think we need to over elaborate on this, it'll likely just confuse people.

Fair enough. It just confused me until I realised what was going on. I agree that it's not relevant to most people, and confusing to put it in the docs for Tree.nodes(), but wondered if it might be worth mentioning in the missing data section?

Will put in a PR for the python version of is_isolated().

[jeromekelleher]

It'd be good to clarify in the missing data section all right, and perhaps give a recipe for leaving out these samples using the is_isolated method.