Thursday, February 24, 2011

You are in a maze of twisty little passages, all alike.

I poked at @rdmpage the other day pointing out that he should think about a 3D phylogenetic tree browser rather than thinking in 2D. Navigating 3D environments is very well worked out thanks to the multi-billion dollar gaming industry, it seems like one could borrow some of the concepts there and map them to viewing phylogenies.

After a little back and forth I decided to do a brain dump based on some ideas that were discussed at the recent GMOD hackathon about linking phylogenies to genome tracks. @hlapp chimed in that this might lead to a career in phylogame (tm?) programming. This resulted in the exercise taking a little more of a fantasy-role than first anticipated, but I think some of the original ideas might still come through. I don't claim that this solves Rod's problem (nor perhaps any real problem), but nevertheless here it is, an A-Z list of possible features for a 3D tree browser using a 3D-game-engine. Such a world might also act as a sort of memory palace. The major thing that's missing- some way to earn "points", because that's what people do in games, get points, so they must be important. If you have to ask- no I do not have a World of Warcraft account, I did however, beat Akalabeth. The graphic below in no way attempts to represent actual phylogenetic relationships, click to enlarge it. If someone wants the illustrator file to modify I'll gladly pass it along.


A - The world consists of a series of paths based on the tree(s) in question. Visible paths are rendered in the ground plane. As the user moves along the paths towards the horizon we see more paths, and the features on the horizon are updated to reflect the remaining tips of the tree.
B - In the horizon pane we see a generalized summary of the clades (paths) we might follow if we continue forward on that path. Much meta-data can be mapped to these features, and the surrounding environment.
C - Bigger features on the horizon (trees?) indicate more paths, giving the user a sense of where the additional information content is.
D - Trees can differ in their health (color of leaves), healthier trees are those that have clades that are supported by more data, or perhaps higher support values.
E - A forest of features might indicate some measure of complexity of the clades in question, or maybe the total number of clades.
F - The weather on the horizon (or above), indicates controversy with the clades in question, or perhaps uncertainty (clouds = doubt).
G - Horizon features need not be trees, they might vary based on some parameter- mountains are rock solid clades.
H - Like the original TOL tree houses you could stop in to learn more about the clades/taxa in question.
I - Trusty map radar.
J - Glowing purple portals let you exit to other websites with context-dependent information.
K - Are there genomes for the clade in question? Lamps, breadcrumbs or other features illuminate parallel paths.
L - Pubs indicated nodes where social-media are discussing the clade in question, pop in for a virtual drink.
M - People are authors, a crowd at a node corresponds to many unique authors.
N - Click to interact.
O - Lateral planes could alternatively map to other metadata. In a 2D system (moving only forward and back) looking left or right could bring up charts/graphs, images or other metadata.
P - Paths could turn to springs, then water courses, for aquatic clades.
Q - For the kids- BATTLE DINOSAURS!
R - Author avatars greet you at the trees they have published.
S - Synapomorphies are statues, monuments or other features along the road. Examine them to learn more.
T - Many trees are contradicted in places, swamps join clades to illustrate alternative arrangements.
U - Fruit on your tree? Someone has left comment(s) or feedback.
V - Strange auras? Multiple equally optimal trees exist.
W - Use the magic wand to toggle through things with auras.
X - Paths can be paved based on the underlying data supporting that clade (gold = morphology, cobblestone = single copy nuclear genes, mud = CO1).
Y - Paths to terminal nodes end with a picture .
Z - Look up! Constellations form alternative phylogenies.

Thanks to PhyloPic.

5 comments:

Rod Page said...

Fun, but I don't think 3D is the way to go. See http://iphylo.blogspot.com/2011/02/why-3d-phylogeny-viewers-don-work.html for some comments.

tjv said...

This is more than fun - it's cognitive liberation!

Hilmar said...

Put this on an iPad as a game and I'd pay to play it with my children.

jhelmt said...

Related: http://fold.it/

Arlin said...

I'd suggest 2 ways to get around the basic problem that the market for tree consumption is way way too small to justify the development effort for a game software focused on navigating a tree.

One approach is just to take the tree aspect out of the foreground, and put it in the background somewhere. Don't make the tree the focus of the game, just make it part of the background knowledge.

People who play dungeons and dragons learn all these arcane rules about different tools and what they do. The idea is just to build a world where the arcane rules are based on actual biology and phylogeny. In the future, I predict, this is the ONLY way we will be able to get kids to learn, because all education will be edu-tainment, and the standards will be REALLY really high.

Apropos, the obvious link with the whole role-playing-game world is that different critters on the tree of life have different skills & tools, right? Do you need some porcupine quills for this adventure, or would you rather have some skunk spray in your bag of tricks? Those are the defensive tools, but maybe you want some offensive tools, like the ability to squirt hot acid (bonus points for knowing where to find that!). Or do you need the more basic ability to generate fuel from sunlight?

It takes knowledge to know where to look for those things. If you put them in rooms and sub-rooms of a castle, where the organization corresponds to taxonomy, you'll be teaching taxonomy. By the way, taxonomy is an indexing system, more for naming and memorization, while phylogeny is a model, more for computing. I don't see a big value to getting people to memorize phylogeny as distinct from taxonomy.

The other approach (the problem of an ultra-small market) would be to generalize the problem of making an educational game that integrates some area of knowledge that is tree-structured. But I have run out of time to describe that idea.