Add supplementary section comparing to other libraries

[jeromekelleher]

This is addressing this reviewer point:

Related to the prior point, it would be helpful to see more detail on comparison to related work, even if just in the supplement, to back up the claims of superiority of tskit to alternatives in the literature with respect to functionality, efficiency, or scalability.

I don't think we need extensive benchmarking here, and a simple table may suffice. We could have some categories like (with better names)

• Large API
• Large objects/trees/graphs/something
• ARG support
• Large ecosystem (supports lots of other tools, rather than one lab's software)
• Languages (C/Python/etc)

Then compare tskit with (say) DendroPy, ARGneedle-lib, and matUtils.

[molpopgen]

We can also note that some (many/all?) of those export/import to/from tskit. I know ARGneedle-lib does. Not sure about the others.

[jeromekelleher]

Could add more specific functionality classes instead of "Large API", like tree traversals, manipulation, popgen stats etc. if we make the rows of the table the categories and the columns the libraries it can be quite a rich comparison.

[jeromekelleher]

We can have a corresponding supplementary section in which we explain the methodology, and list out examples of the types of operation in each of the classes.

[hyanwong]

Do we mention April's GRG format?

[hyanwong]

It wasn't mentioned in this specific reviewer comment, but it might be worth indicating in the comparison table whether there is extensive documentation & tutorials etc. I think this is key to library use, but I don't know how to make this a fair comparison. DendroPy would score very highly here, of course (which is good).

[jeromekelleher]

Do we mention April's GRG format?

Good idea - if it fits in naturally then I think it's a good addition. However, it might be a bit awkward to explain the distinction between the GRG and ARGs