Treemachine could manage multiple trees

[jar398]

We've been talking about this for a long time, and I think there has been some implementation work, but I can't find any issue for it, so creating one now.

I'd be interested in hearing the use cases for this feature before going much farther with it.

[bomeara]

Reproducibility. "We used the synthesis tree version 2". "Sorry, all available now is version 5"

[arlin]

Phylotastic has a use-case for maintaining metadata on multiple synth trees. We want clients in the phylotastic system to provide provenance information to users. For workflows that depend on subtree or induced_subtree, this means (among other things) getting the synth tree version.

Over time, as the phylotastic system gets used, the world becomes populated with trees with metadata records that say "derived by induced_subtree from OT synth tree version X", where X is 8, 9, 10, 11 and so on.

At the least, I think, OT needs to make the metadata for old tree versions available. This does not necessarily mean programmable access-- it could be a web page with a dump of the "about" response for each version.

There is then, in principle, the case where someone in 2017 wants to go back and retrace their steps from a year earlier and execute induced_subtree with taxa A, B, C, and D using synth_id = 8. But that is going to be a rare case.

[arlin]

A more interesting case would be using the availability of separate versions to study the improvement in quality of the synth tree.

[mtholder]

Just filling in some of the gitter chat starting about https://gitter.im/OpenTreeOfLife/public?at=585adfd1058ca96737771f8b
I think we will be able to provide API access to the metadata for different synth runs.
The memory/storage requirements for storing a large number trees could be problematic, and the workaround for not finding a hosted tree will be to check the static uploads using https://tree.opentreeoflife.org/about/progress to find versions by date (until we have that info served up in some web service method)