from_ms: assign samples to populations

[grahamgower]

Currently all samples are assigned to a single population. But the ms command line is included in the ms output, so the sampling configuration could be parsed. This could be used to assign the samples to the correct populations.

[jeromekelleher]

I guess we could use demes to parse the ms command line?

[grahamgower]

We only need to parse one option (-I npop n1 n2 n3 ... [4Nm]). If it's present, use it, otherwise there's only one population. I think pulling in demes here would be overkill.

[grahamgower]

Hmm, actually, this could be more complicated if the user did temporal sampling with -eA. Probably easiest to just detect that and throw an error.

[jeromekelleher]

OK - given the fundamental limitation on what we can read in from ms with the branch lengths though, I'm pretty lukewarm on putting effort into this.

[grahamgower]

Sure. My specific use case would be better served by adding an option to mspms to output a tree sequence anyhow. :)

[jeromekelleher]

Hah, yes, good point. It's quite a roundabout way of getting a tree sequence out of msprime!