Participants

Joe Felsenstein, Sheldon McKay, Brian O'Meara, Natalie Henriques, Liya Wang

Action items

• Brian will send Liya info on ancestral state reconstruction intros (lectures, books, etc.) [sent references to textbooks by Felsenstein and Yang]

Agenda

Agenda

• review action items from previous meeting
• review progress for WG, project goals and timelines
• what does iPlant trait evolution success look like in 6 months' time?
• open discussion

Minutes (links to action items in bold)

Problems connecting at first due to change in WebEx protocol (it has been updated on the wiki)

Some discussion about easy implementations for ancestral state reconstruction of continuous traits: perhaps wrapping ANCML (a fork of CONTRAST) or using a different published algorithm (Liya will forward - A4). Some discussion about who would do the implementation (is it a core software thing or something that would be done by engagement team or someone at TACC?) and feasibility of putting it in release 1.0 vs a future release. Conclusion seemed to be that adding features for this release is unlikely.

Progress on software: file upload interface is ready (though not viewable yet), wrapping of Contrast will happen in January for Feb. release.

How important is collaboration (i.e., sharing data within the discovery environment): conclusion is that it's not very important for this group (though having this ability is not objectionable). However, some later discussion about tree-sharing (below).

Discussion of what success will look like in 6 months (i.e., 4 months after the next release). It will mean inclusion of at least our top three analysis methods and some ways to visualize info (like reconstructions) on a tree

Joe brought up the issue of combining multiple trees, especially from trees uploaded to the discovery environment. Current plan is to have user trees at first, then big iPToL tree later, but what about already existing trees (published and in progress). It would be helpful to have these available to use (i.e., to ecologists with trait data but no trees). A2

Also discussion of what sort of branch lengths the big tree will have. (A1)

Some discussion about whether to visualize contrasts on tree. Joe warned that contrasts (without rerooting on each node to get the best estimates) are not that informative. Standardized contrasts can be used to get "local" rate estimates (see Garland. Rate Tests for Phenotypic Evolution Using Phylogenetically Independent Contrasts. Am. Nat. (1992) vol. 140 (3) pp. 509-519), though adding a rerooting step for each of these might be better (and see O'Meara et al. Testing for different rates of continuous trait evolution using likelihood. Evolution (2006) vol. 60 (5) pp. 922-933).

Brian noted that many of the issues about uses besides the one user+one tree+one set of tip data model were brought up back in August (see here ), and it might be worth revisiting these.

New action items

A1: Sheldon will coordinate with big trees group about whether their tree will have branch lengths in units of amount of change and/or calibrated to time (having both would be good)

A2: Sheldon will investigate availability of existing large trees within discovery enviroment

A3: Joe may start developing a set of criteria about when users should be especially worried about the appropriateness of the methods

A4: Liya will send reference about algorithm for inferring ancestral states using PIC info (already done: Garland et al. An introduction to phylogenetically based statistical methods, with a new method for confidence intervals on ancestral values. Am Zool (1999) vol. 39 (2) pp. 374-388)