-
Notifications
You must be signed in to change notification settings - Fork 3
Design notes
We want an illustration tool that supports these goals:
- dynamic, data-driven illustration
- print-ready output (SVG => PDF, and possibly interactive versions)
- repeatable process (e.g., re-render after changes to data)
- shared expertise and "house styles" for journals or institutions
- rich possibilities for annotation and ornaments
- zero install, easy to get started
Our starting point is to think in terms of a decision cascade and rendering pipeline, as illustrated below (the editable doc is available on Google Drive):
This design separates decisions made at different times, by different users, and supports sharing and re-use. It sidesteps many important choices about data storage, social/sharing features, ownership and permissions. Our initial proof-of-concept tests will focus on the usability and output of this model with mocked data sources as input.
web-based -- runs in modern browsers (based on support for d3/svg)
separation of “content” (the tree as a data structure and its associated information) and “layout/style” -- how the information should be rendered. E.g.
- rectangular vs circular layout of nodes/branches
- show branch support as numbers at nodes vs thickened branches over threshold
- render tip labels as taxon names (given mapping)
- more general mapping of selected data/attributes to styles?
- to support, eg, rich edges (coded by color, thickness, linestyle)
figures can be saved as vector graphics (SVG)
- can the phylogenetic info be embedded in (saved with) the SVG, so that it could be round-tripped?
-
or should SVG be treated as ephemeral, with separated tree data + style?
- supports full round-tripping with updated external tree (as Newick, NeXML, …)
options for interactivity:
- zooming, panning
- “collapsible” clades
- “replay” (step through a series of zooms, pans, expanding/collapsing, …)
sharing styles
- browsable gallery of shared+named style settings
- show each applied to a “generic” example tree
- quickly choose and apply chosen style to your tree
- known-good styles, eg, “Radial layout with expressive links”
- “house styles” tailored for particular journals/venues
- preferred typefaces
- minimal line-weight at final printed size
- eg, “Systematic Biology (print, b/w line art)”
- eg, “FigShare slideshow w/ captions”
the BIG TREE problem and the need for a client-server API
- partial/selective rendering needed for big trees
- cannot rely on whole tree to fit into client’s memory
- need protocol for client to:
- ask a server about resources (trees)
- make (stateful?) requests to the server
integrate this into the Open Tree curation app, in Tools tab?
- easy selection of any tree in the current study
- choose and apply settings from Shared Styles
- link to full tree illustrator in a separate window?
- what is the result? maybe just facilitates opening this study’s tree(s) in the illustrator app?
usable from scripting languages = important [imho - RR]
some notes from the F2F meeting of larger Open Tree team [jimA]
- feature wish list
- collapsing clades
- highlighting clades
- coloring clades
- adding images
- COMPLETE SOLUTION w/o needing other graphics tools (survey question: "What do you use Illustrator for?")
- We absolutely need to interview a set of experts, ie, people who make nice tree figures “the hard way” through a more difficult workflow
- look for common feature set among the (many) existing tree-illustration tools
- focus on OCCASIONAL users, who need an easy-to-use tool
- look at possible interop INSTEAD of a major new tool?
- use one or more of the (many) existing tree-graphic tools
- help to modify/style output from these?
- offer option to label tree with original OR mapped taxon names?
- SPIRITED DISCUSSION! about whether journals have tricky formatting requirements for tree illustrations. What's the real deal?
- To elicit contributions, our value proposition is probably not The Ultimate Drawing Tool, more like the Path of Least Resistance:
- dynamic (data-driven)
- capture styles and illustration-specific decisions
- really easy to use
- good enough for publication