Very cool. Perhaps you've seen OneZoom[0] already which is purely based on phylogenetic data sourced from Open Tree of Life. OTOL has been around for a long time. The project aggregates phylogenetic data from thousands of studies and creates a massive synthetic tree.[1] It's neat because anyone can contribute. If you find a study not represented in there you can contribute the newick tree/annotation for it.
OneZoom is a great museum piece. Very engaging visualization that invites large touchscreen interaction. Impressive code behind it too for pruning offscreen vectors so the zoom stays smooth. On desktop however, perhaps due to my age, it gives me a bit of gorilla arm.
Thanks to open data, I offer the same node hierarchy of OTT, among 10 other backbones, but present it with 3 different (perhaps less dazzling) viewing mechanics switchable at the top: Icicle, which packs hierarchy into a fixed vertical height while allowing depth to stay horizontally reachable. Chronogram is the opposite, showing structure even 10 levels deep in a compact horizontal region, but at the expense of large vertical expansion. And Tree, the default, which is a compromise that can also show multiple winding paths simultaneously by folding them against the gutter.
I also focus less on the tree itself and more on trait overlay and rolling up thousands or even millions of species into compact segments so you can get a feel for how many chordates lay eggs or how many spiders are carnivorous at a glance.
https://ptree.org/#OTT
Super cool! I just learned about 'incertae sedis' from here—though I think these things should at least be in their kingdom (e.g. Naldaviricetes are clearly viruses no?)
Absolutely agree with the underlying UX point. "Incertae sedis" shouldn't leave you unable to tell what broad kind of organism you're looking at.
I do filter structural junk per backbone, but only from evidence that backbone itself supplies: synonym/accepted status, "not-a-taxon", environmental, hidden, unplaced flags. I also insert neutral buckets such as "classes (no phylum)" where ranks are skipped, so thousands of taxa don't spill into one flat level.
What I deliberately don't do is move a taxon to the parent that I think makes more biological sense. Ptree (Ptable gets the same kind of requests to classify metalloids) is downstream of academic taxonomic authorities operating under something like Wikipedia's "no original research" policy. It can expose disagreement among authorities, but it shouldn't quietly become another undocumented taxonomy.
Naldaviricetes is a good example. It is unquestionably a class of viruses, but "Viruses" is not necessarily its formal kingdom. ICTV currently leaves it unassigned to a realm, kingdom, or phylum so Ptree's ICTV backbone places it directly beneath a display root called Viruses. NCBI places it beneath "Viruses incertae sedis," while GBIF supplies it without that viral higher lineage, producing the confusing placement you found.
The tree map folding is ingenious. I have never seen it.
As a whole it's brilliant and humbling. Both the vastness and complexity of nature and biology on Earth, and also the absurd amount of work and human lifetime that has been spent on gathering the information. It's also a testament of the power of Wikipedia.
The problem with rolling up 2.2 million species into 7 kingdoms is it surfaces data quality issues loudly. Ranking photos by popularity means one node mismatch in millions due to a homonym will send that up from 10 levels deep. As I find these, I'm trying to squash them.
That raccoon you see is actually suffering from Morbillivirus Canine Distemper Virus and that photo is correctly attached to that virus, and that bird is indeed eating a lungfish. I actually do penalize outlier favorite counts for photos because they're often meme-ish. I may blacklist this so the 2nd most favorited lungfish can surface. There are a lot of detailed rules that score which photo you get based on aspect ratio, whether the subject is in captivity, alive, etc.
Love this. Also love that this is coming from someone with expertise who, based on comments below, takes quality seriously. Particularly cool when juxtaposed against one the other day “Any Human Ever” which was at first enthralling and joy-inducing, but upon deeper inspection was just a slop factory spreading misinformation across the internet.
This is the highest compliment, and I see why no other projects have tried to join all the different trees of life together or attach and reconcile traits. This was the majority of the project.
However, the unique "roll-up" visualization of traits brought in an interesting method of self-correction. For example, if any other phyla than the obvious 4 showed up with a single egg-laying species, I went back to the drawing board to find out where that incorrect join introduced the error.
[0] https://www.onezoom.org/
[1] https://tree.opentreeoflife.org/opentree/argus/ottol@378964/...
Thanks to open data, I offer the same node hierarchy of OTT, among 10 other backbones, but present it with 3 different (perhaps less dazzling) viewing mechanics switchable at the top: Icicle, which packs hierarchy into a fixed vertical height while allowing depth to stay horizontally reachable. Chronogram is the opposite, showing structure even 10 levels deep in a compact horizontal region, but at the expense of large vertical expansion. And Tree, the default, which is a compromise that can also show multiple winding paths simultaneously by folding them against the gutter.
I also focus less on the tree itself and more on trait overlay and rolling up thousands or even millions of species into compact segments so you can get a feel for how many chordates lay eggs or how many spiders are carnivorous at a glance. https://ptree.org/#OTT
I do filter structural junk per backbone, but only from evidence that backbone itself supplies: synonym/accepted status, "not-a-taxon", environmental, hidden, unplaced flags. I also insert neutral buckets such as "classes (no phylum)" where ranks are skipped, so thousands of taxa don't spill into one flat level.
What I deliberately don't do is move a taxon to the parent that I think makes more biological sense. Ptree (Ptable gets the same kind of requests to classify metalloids) is downstream of academic taxonomic authorities operating under something like Wikipedia's "no original research" policy. It can expose disagreement among authorities, but it shouldn't quietly become another undocumented taxonomy.
Naldaviricetes is a good example. It is unquestionably a class of viruses, but "Viruses" is not necessarily its formal kingdom. ICTV currently leaves it unassigned to a realm, kingdom, or phylum so Ptree's ICTV backbone places it directly beneath a display root called Viruses. NCBI places it beneath "Viruses incertae sedis," while GBIF supplies it without that viral higher lineage, producing the confusing placement you found.
Discussed a few times over the years!
Interactive Periodic Table - https://news.ycombinator.com/item?id=27650576 - June 2021 (22 comments)
Dynamic Periodic Table (2017) - https://news.ycombinator.com/item?id=20495105 - July 2019 (21 comments)
Worlds Best Periodic Table - https://news.ycombinator.com/item?id=4378346 - Aug 2012 (5 comments)
Dynamic Periodic Table - https://news.ycombinator.com/item?id=3199922 - Nov 2011 (20 comments)
Dynamic/Interactive Periodic Table - https://news.ycombinator.com/item?id=1193431 - March 2010 (2 comments)
As a whole it's brilliant and humbling. Both the vastness and complexity of nature and biology on Earth, and also the absurd amount of work and human lifetime that has been spent on gathering the information. It's also a testament of the power of Wikipedia.
The project only exists thanks to thousands of scientists carefully cataloguing all this data and releasing it CC-BY. I am only the interface.
That raccoon you see is actually suffering from Morbillivirus Canine Distemper Virus and that photo is correctly attached to that virus, and that bird is indeed eating a lungfish. I actually do penalize outlier favorite counts for photos because they're often meme-ish. I may blacklist this so the 2nd most favorited lungfish can surface. There are a lot of detailed rules that score which photo you get based on aspect ratio, whether the subject is in captivity, alive, etc.
However, the unique "roll-up" visualization of traits brought in an interesting method of self-correction. For example, if any other phyla than the obvious 4 showed up with a single egg-laying species, I went back to the drawing board to find out where that incorrect join introduced the error.