Tardigraph
Start here

Your first contribution

Four things you could finish today. Each one is written as numbered steps, none of them needs an account or permission, and the smallest is genuinely useful — we are not inventing busywork to make you feel included. Do one step and stop if you want to. Partial is fine; partial is how most of this project got built.

What we promise in return. Anything you send that we use is credited to you by name on the page or in the paper it touches, unless you ask us not to. You will get a reply from a person. And if what you send turns out to be wrong, we will tell you what we found and why — that is worth having too, and several of our own findings have died that way.

1. Find a tardigrade and tell us where

needs: moss, a jar, and any microscope — a $30 one works

  1. Take a handful of moss or lichen. A roof, a wall, a tree trunk, a gutter. Dry and crusty is better than lush and green.
  2. Soak it in water for two to twenty-four hours in a shallow dish. Rainwater or bottled is better than chlorinated tap water.
  3. Squeeze the moss out over the dish and throw the moss away. What you want is the water at the bottom.
  4. Let it settle for an hour, then pipette from the very bottom onto a slide or into a shallow lid.
  5. Look at 40× or so. Tardigrades lumber. If it moves like a tiny bear rather than darting like a rotifer, that is one.
  6. Photograph it however you can — a phone held to the eyepiece is completely fine and is how most records are made.

Send us: the photo, roughly where you were (a suburb is enough — no need for an exact address), the date, and what the moss was growing on. That is a record.

Why this is worth doing. Distribution data for most tardigrade species is thin, patchy and old. Several of the papers we filed this week are exactly this: somebody looked at moss in a place nobody had looked, and wrote down what was in it. There is no minimum qualification for that, and there is a longer guide here if you catch the bug.

2. Check one claim against its source

needs: twenty minutes and the ability to read carefully

  1. Open the knowledge graph and pick any claim that interests you. Every one names the paper, and the page or figure, it came from.
  2. Find that paper. Many are open access and linked. If it is paywalled, pick a different one — we are not asking you to buy anything.
  3. Read the sentence we cited and ask one question: does the paper actually say what we say it says? Not "is it true" — "did we report it correctly".
  4. Write down what you checked, even if it was right. A claim confirmed by somebody who is not us is worth more than one we checked ourselves.

Send us: the claim id, and either "checks out" or what is wrong with it. One line is plenty.

Why this is worth doing. We have 1,690 claims and two people. We have already caught ourselves misattributing a paper, giving a wrong DOI, and stating an hour that was half of what it should have been. Everything here is built on the assumption that the citations are right, and nobody outside this project has ever tested one.

3. Give an unnamed protein a name

needs: a terminal, and a willingness to paste two commands

  1. Clone the repository and get one of the loci we have not identified — there are thirteen from the last two days alone, listed in docs/IDEAS.md.
  2. Pull its protein sequence out of the genome database: 
    python3 tools/tg.py show gene:<id>
  3. Put the sequence into any tool you trust — InterPro, HHpred, Foldseek, a BLAST against SwissProt. You are looking for what family it belongs to.
  4. Say how confident you are and why. "Domain architecture matches X" is a different claim from "best hit was X at E=1e-40", and we record which one it is.

Send us: the locus, what you think it is, which tool said so, and the numbers. We will put it in the graph with your name on the identification.

Why this is worth doing. Two thirds of this genome is annotated only "hypothetical protein". Every name is a small permanent addition, and a wrong name that gets corrected later is still progress — we keep both, with the correction attached.

4. Tell us something here is wrong

needs: five minutes and mild irritation

  1. Read any page on this site.
  2. Notice something that is wrong, unclear, broken or badly explained. A typo counts. A confusing sentence counts. A chart you could not read counts. "I did not understand what this page was for" is one of the most useful things you can say.
  3. Tell us the page and the thing.

Send us: the URL and what went wrong. No format required.

Why this is worth doing. The front page said an animal was 152 millimetres long for a week, because a stylesheet turned a µ into an M and nobody looked. Fresh eyes catch what familiar ones cannot.

Sending it

Everything above goes to one place: meow@meow-meow.io. Put whatever you like in the subject line. There is no template, no form, no account, and no contributor agreement — there is a person who reads it and writes back.

If you would rather do something bigger, the longer list of what needs doing has the jobs that are actually blocking things.