Viewer · a mode
FoldSpotter
A field card, like bird spotting, but the bird is a protein. Pick a square, then pick the residue on the chain where you think it is. One square on every card is not in this protein: say so when you find it.
Prediction Computed by a model, with how sure it is. Not an experiment.
Choose a protein:
Pick a square first.
Under each letter: H very high (90+) · C confident (70–90) · L low (50–70) · V very low (below 50), with AlphaFold's colours.
How every square is checked
- An α-helix: a stretch AFDB's secondary-structure assignment (DSSP) calls an α-helix, 7 residues or longer.
- A β-strand: a stretch DSSP calls a β-strand (part of a sheet), 3 residues or longer.
- A loop: 6 or more residues in a row that are neither helix nor strand.
- A tight 3₁₀ helix: a short, tighter helix (three residues per turn) that DSSP marks separately.
- A floppy end: an end of the chain where 10 or more residues in a row are below pLDDT 50.
- A very sure stretch: 20 or more residues in a row at pLDDT 90 or above.
- An unsure middle: 5 or more residues in a row below pLDDT 70 that touch neither end.
- A disulfide pair: two cysteines whose sulphur atoms sit within 2.5 Å in the model: one residue of the pair counts.
- A proline: a proline residue (letter P), which bends the backbone.
- A ring side chain: phenylalanine, tryptophan or tyrosine (F, W, Y): side chains with a flat ring.
- A glycine: a glycine residue (letter G), the only amino acid with no side chain.
Helix and strand come from the secondary-structure assignment (DSSP) that the AlphaFold Database ships inside each model file. Confidence is the model's own pLDDT. A disulfide is drawn when two cysteine sulphurs sit within 2.5 Å in the model. All of this describes a prediction, not an experiment: the AlphaFold DB model is one chain on its own, so nothing that binds it (partners, metals, haem) is there.
Models: AlphaFold DB (CC BY 4.0), snapshot 2026-10-04.