PencilOrInk · a mode
FirstLook
Deal one protein. Turn it. Draw it: ink where AlphaFold is sure, pencil where it is not. Give it a name. Then file your field note in your Bestiary.
1. Deal
The pool holds 53 proteins. Each one has a real AlphaFold model.
No protein dealt yet.
Same protein for the whole room: put the deal number in the address, like
/tools/pencilorink/firstlook#deal=12. Every screen with that address gets the same protein.
2. Turn it
Turn the model with a finger or a mouse. Look at it from the side and from the end.
3. Draw it
This strip is the chain, start to end, one mark per amino acid. Draw in ink where the strip is ink. Draw in pencil where the strip is pencil.
ink: AlphaFold's confidence is 70 or more pencil: below 70
4. Name it, 5. file it
Give it your own name on the field note. Write what you saw in one column and what you think in the other. Then put the sheet in your Bestiary.
the Sketcher's drawing Prediction Computed by a model, with how sure it is. Not an experiment.
Field note · FirstLook
| Protein | ________________________ |
|---|---|
| UniProt | __________ |
| How sure | ________________________ |
A prediction, not an experiment.
What I saw
What I think
My name for it: ____________________________ Date: __________
For the grown-up at the table
- One screen is enough, or none: print field notes ahead and deal by reading out a deal number.
- Anyone can leave early. Nothing breaks and nothing is lost, because nothing is kept here.
- Nothing is saved on this site. The Bestiary is a paper binder. The children keep it, not us.
- No accounts, no typing, no scores, no streaks.
- "What I saw" is for what is on the screen or the strip. "What I think" is for guesses. Keeping them apart is the point of a field note.
All 53 proteins in the pool
- ABL1 kinase (P00519)
- EGFR (P00533)
- RuBisCO (P00875)
- Ovalbumin (P01012)
- Growth hormone (P01241)
- Insulin (P01308)
- Antibody (IgG) (P01857)
- Myoglobin (P02144)
- Collagen (P02452)
- Alpha-crystallin A (P02489)
- Casein (P02662)
- Beta-lactoglobulin (P02754)
- Serum albumin (P02768)
- Lactoferrin (P02788)
- Ferritin (P02794)
- Hemagglutinin (H1N1 flu) (P03452)
- Antifreeze protein (type I) (P04002)
- Catalase (P04040)
- Aldolase A (P04075)
- Keratin (P04264)
- p53 (P04637)
- Sodium-potassium pump (P05023)
- ATP synthase (P06576)
- Trypsin (P07477)
- Rhodopsin (P08100)
- Firefly luciferase (P08659)
- Lactase (P09848)
- DNA polymerase (P09884)
- Pepsin (P0DJD7)
- Calmodulin (P0DP23)
- SARS-CoV-2 spike (P0DTC2)
- Alpha-amylase (P0DUB6)
- Myosin (P12882)
- CFTR (P13569)
- Dopamine D2 receptor (P14416)
- GLUT4 (glucose transporter) (P14672)
- Tyrosinase (P14679)
- Hexokinase (P19367)
- Acetylcholinesterase (P22303)
- RNA polymerase II (P24928)
- Aquaporin-1 (P29972)
- BRCA1 (P38398)
- Green Fluorescent Protein (P42212)
- Lysozyme (P61626)
- Ribosomal protein S6 (P62753)
- Actin (P68133)
- Histone H3 (P68431)
- Hemoglobin (beta) (P68871)
- Hemoglobin (alpha) (P69905)
- Cytochrome c (P99999)
- Tubulin (Q71U36)
- CRISPR Cas9 (Q99ZW2)
- ACE2 (Q9BYF1)
Honest notes
- AlphaFold gives every amino acid a confidence score from 0 to 100 (pLDDT). 90 and up is very high, 70 to 90 is confident, 50 to 70 is low, below 50 is very low.
- Ink at 70 or more is our drawing rule, not AlphaFold's: it puts "very high" and "confident" in ink.
- Where AlphaFold is unsure, the chain is often floppy in real life, with no one fixed shape. Other times the model had too little to go on.
- The pool is every protein our site already names in plain words that has an AlphaFold model we keep a copy of. The scores come from our copy of the AlphaFold Database.
Sources
- Jumper et al. 2021, "Highly accurate protein structure prediction with AlphaFold", Nature: doi.org/10.1038/s41586-021-03819-2
- Tunyasuvunakool et al. 2021, "Highly accurate protein structure prediction for the human proteome", Nature (low confidence and disorder): doi.org/10.1038/s41586-021-03828-1
- AlphaFold Protein Structure Database, FAQ (the confidence bands): alphafold.ebi.ac.uk/faq