AlphaFold Viewer
Also: FoldSpotter, a field card: find a helix, a strand and a floppy end on a real model.
A free, zero-install viewer for the AlphaFold Protein Structure Database. Search by protein name or UniProt accession, or pick a protein below. Structures are coloured by pLDDT confidence — the same honest confidence signal FoldLens uses. Nothing is installed; nothing about you is collected.
Also: TourMaker, make a short guided tour of AlphaFold structures that lives in a link.
Pinch to zoom · drag to rotate
Keyboard shortcuts
Click the structure (or Tab to it) first, then:
- R reset the camera to the default view
- S start / stop slow rotation
- F enter / exit full screen
- C show / hide the Mol* control panel
Rotate, pan and zoom with the mouse, trackpad or touch.
Predicted aligned error (PAE) shows AlphaFold's expected position error, in ångströms, between every pair of residues. Dark regions mark residue pairs whose relative position the model is confident about; bright regions mark pairs whose relative arrangement is uncertain — often between two separately-confident domains. It complements pLDDT (per-residue confidence) with pairwise, inter-domain confidence. PAE data (JSON)
Story
What the confidence colours mean
- Very high (pLDDT ≥ 90)
- backbone and side chains are typically both predicted with high accuracy.
- Confident (90 > pLDDT ≥ 70)
- the backbone is usually modelled correctly; some side chains may be misplaced.
- Low (70 > pLDDT ≥ 50)
- treat with caution — often flexible or intrinsically disordered regions.
- Very low (pLDDT < 50)
- should not be interpreted — typically disorder, or regions the model cannot confidently place.
A high pLDDT for every domain does not mean AlphaFold is confident in how those domains are positioned relative to one another — that pairwise, inter-domain confidence is what the Predicted Aligned Error (PAE) shows.
Model: AlphaFold DB · — · CC BY 4.0 · prediction, not experiment.
Caption for this model
Written from the model's own record, never typed: its name, the model version and date, how many residues fall in each confidence band, and what kind of picture it is. Save figure and Print as poster carry it too.
Sources: the model and its per-residue confidence file are from the AlphaFold Database (Varadi et al., Nucleic Acids Research 2024, doi:10.1093/nar/gkad1011); the four confidence bands are AlphaFold's own (Jumper et al., Nature 2021, doi:10.1038/s41586-021-03819-2); the current sequence and its version are UniProt's (UniProt: entry and sequence versions).
PrintFold: a model you can hold
Save this model as a file for a 3-D printer. The file is a tube that follows the protein chain. How thick the tube is shows how sure AlphaFold is: thick where AlphaFold is sure, thin where it is unsure, in four steps (very high, confident, low, very low). So you can feel the confidence with your fingers, in the line of HearTheChain, which lets you hear it.
A prediction, not an experiment. Thin = AlphaFold is unsure there; it is not a measured flexibility. The file is made in your browser from the AlphaFold DB model file; nothing is uploaded. Thin parts may need supports when printing.
Why the colours you just read matter
On the default protein, p53, the confident middle is the DNA-binding domain, where most of the cancer-linked changes in the gene fall, while the pale, low-confidence ends are flexible in the real protein too. Whatever protein you open, read the colours as the model's own confidence, residue by residue, and the PAE map for how its parts sit relative to each other.
About this viewer
Models come straight from the AlphaFold Database (EMBL-EBI / Google DeepMind) and are rendered in your browser with Mol*, the open-source structural-biology viewer, via PDBe Mol*. Both are self-hosted here — no third-party CDN, no tracking. AlphaFold predictions are computational models, not experimental structures; the pLDDT colouring shows where the model is confident (blue) and where it is not (orange).
AlphaFold DB data is released by EMBL-EBI under CC BY 4.0.
Please cite the AlphaFold papers when you use a structure — see
the AlphaFold DB FAQ. This viewer is part of the
free, non-profit Fold Commons project; you can cite the tool itself via the
repository's CITATION.cff.