Tree Visualization Workshop

The iPlant Tree of Life (iPToL) Tree Visualization Working Group, hosted a workshop at the Texas Advanced Computing Center (TACC), Austin, TX, on February 25 - 26, 2010.

Executive Summary

The purpose of the workshop was to bring together plant scientists and computer scientists to advance the design of tree visualization tools in order to support current and future needs. The key outcomes are as follows:

The basic design of the Big Tree prototype was validated; concrete, prioritized features were identified for its development
The tree visualizer will be a stand-alone component that will accept metadata from other programs and applications.
Metadata about gene expression, traits, etc, will be mapped onto the nodes, edges, clade triangles, and text in the tree visualization
The prototype will serve as the first tree visualization deliverable to the external community; a final version will be developed based on this design
We identified several needs for visualizing multiple phylogenies. We will focus on two:
- reconciled gene trees and species trees, directed by the needs of the Tree Reconciliation working group
- methods for visualizing thousands of input trees, which is a much more open research question

Discussion

The major topic of the meeting was to present, critique, and advance the design of the Big Tree prototype, which has met a major milestone by providing an interactive visualization of million tip large phylogenetic trees. The team provided detailed comments, questions and critiques on the prototype, and explored specific design issues, such as branch layout, text and labeling, scalability, fluidity of browsing, and ease of navigations. Discussion about extensions and additional features for the current prototype followed. Some key topics included what user interaction capabilities to support, such as collapsing levels, rotating clades, adding labels, adding annotation, etc, what analytical operations to include, what additional views to support (e.g., drill downs, alternative representations), and ideas on designing a clear, intuitive, aesthetic user interface. The team also discussed and prioritized also future ideas for the Big Tree viewer, addressing dynamic linking with other visualizations (e.g., a gene browser), drill downs to annotations comparative maps, trait databases, the literature, needs of other iPLANT groups, and the addition of images.

Metadata played a key role in our thinking, including consideration of what types of metadata that could, or should, be mapped onto the tree, and methods for visualization these metadata.

Wrapping up on this topic, we focused on next steps and action items for the Big Trees deliverable and identified key features to incorporate.

The second major topic for the workshop was visualization for tree comparisons. In this use case, the goal is to provide insight into tree reconciliation, co-evolution, mapping gene trees and species trees. We addressed how visualization help in extracting the phylogenetic signal in thousands of input trees, how to integrate the fundamental statistical and analytical methods for this domain, and what methods are needed to identify, annotate and communicate patterns.

The meeting ended with a discussion of scientific problems that will be addressed by the iPToL tree visualization innovations, focusing on what scientists will be able to do that has never been done before. Just providing easy access to all the relationships between a million species will, in itself, bring new insights, for plan biologists and for other scientists looking at data with a phylogenetic linkage. Being able to see various types of metadata will add new insight, enabling scientists to integrate gene expression and trait data into their analysis. Some domains that will benefit are the development of pest and drought resistance, gain better understanding of endangered species, understanding how species co-develop, such as corals and fishes, plants and insets.

Finally, the Tree Visualization working group should now start to work more closely with the other iPToL working groups. For example, the Trait Evolution group requires that the large tree viewer can display output from their analyses, and the needs of Trait Reconcilation should direct where we focus our efforts on tree comparisons. The Data Integration group can help us to define the API that we use to input and output trees and metadata.

Action items for the large tree viewer, metadata and tree comparisons are described on the following pages:

https://pods.iplantcollaborative.org/wiki/display/iptol/Large+tree+wish+list

https://pods.iplantcollaborative.org/wiki/display/iptol/Metadata+wish+list

https://pods.iplantcollaborative.org/wiki/display/iptol/Tree+Comparisons

.