include "load" as a statistic

[petrelharp]

There's a few places where "average number of derived mutations carried by a sample" is a useful statistic; for instance, in @nspope's renormalization, and for @mufernando's quantifying of fixation rate. This is a statistic with summary function lambda x: x and polarised=False. Is load a good name for this?

[nspope]

What about mutational_load -- qualifying "load" somehow seems like a good idea, as I think having a statistics function called TreeSequence.load would be confused with tskit.load.

[petrelharp]

Lol good point. Alternatively, perhaps ts.distance_to_root or something like that?

[gregorgorjanc]

I like distance. The docs can mention that such a stat is sometimes called "load", though I wonder if the later is usually meant with polarised alleles. Should polarisation be an argument?

[jeromekelleher]

perhaps ts.distance_to_root or something like that?

SGTM. Just thinking about the edge cases, that's distance to root measured in units of branch length (branch mode) or mutations (mutation mode)? Does it have an interpretation in site mode?

[petrelharp]

You mean node mode, I assume? Yes, that'd be something like the total amount of genetic material inherited in the samples from the node, I think?

[apragsdale]

I agree that it would be useful to have a statistic describing the number of mutations carried by a sample.

I don't know if it should be called or even described in the docs as "load", because as far as I know, we can't tell from the tree sequence (aside from metadata) whether a mutation is deleterious or neutral. And I don't know that you can get something like expected load from branch mode in any case, since any deleterious mutation that arises will itself distort the genealogies, so by conditioning on the genealogy in branch mode, we get the expected number of neutral mutations carried by samples instead.

[nspope]

Good point about deleterious -- I'd been referring to the "neutral mutation count" as load but that's not what people generally mean by it.