About RataFold

RataFold is a conformational variant atlas for proteins. For each protein, we provide the native fold (from AlphaFold) plus a curated set of alternative conformations — the shapes the same backbone could plausibly adopt.

What this complements

AlphaFold predicts a single most-likely structure per protein. RataFold extends this with a conformational neighborhood — the alternative folds, intermediate states, and lower-energy configurations that exist around the native prediction. Useful for ensemble docking, druggable intermediate identification, and structural variation analysis.

How it works

RataFold uses proprietary conformational search technology to explore the vast space of possible backbone arrangements and surface the most relevant alternatives for each protein. For each target, we report the top-ranked alternative conformations along with their relative energy and structural distance from the native fold.

Honest scope

The alternative conformations provided here are candidate structures intended as starting points for downstream analysis (molecular dynamics validation, docking studies, experimental investigation). RataFold does not replace AlphaFold for single-structure prediction, nor real-physics simulation tools for stability assessment.

Why conformational variants matter

Proteins are not static. A single 3D prediction is a useful starting point, but the same amino-acid chain in a cell adopts many shapes — opening and closing to bind partners, flexing during catalysis, occasionally misfolding. Two foundational ideas:

• Anfinsen's Dogma (1973): the native fold is the global minimum of free energy. So lower-energy variants of the predicted fold are candidates worth examining — the prediction may not be the true minimum.
• Levinthal's Paradox (1969): a 100-residue protein has on the order of 10⁹⁵ possible conformations. Random search is impossible — proteins funnel along energy gradients, and so does any search that hopes to find biologically relevant states.

RataFold surfaces a curated set of low-energy alternatives that may correspond to: binding-competent states, hinge / breathing motions, folding intermediates, allosteric forms, or partially-misfolded precursors of disease states.

Data schema

Each protein record at data/proteins/<PREFIX>/<ID>.json has the following shape. .js twins exist for file:// use.

{
  id:           "P01308",          // UniProt accession
  name:         "Insulin",         // UniProt description
  gene, organism, taxId, length, sequence,

  native: {
    source:     "AlphaFold v6",
    plddtMean:  52.91,             // mean plDDT confidence
    energy:     605.88,            // RataFold force-field score
    ca:         [[x,y,z], …],      // Cα coords (length = residues)
    residues:   [{                 // full-atom per residue
      resSeq:   1,
      resName:  "MET",
      plddt:    64.44,             // per-residue confidence
      atoms:    { N:[x,y,z], CA:…, C:…, O:…, CB:…, … }
    }, …]
  },

  variants: [{                     // one per discovered variant
    rank, energy, stability,
    rmsdNative,                    // overall Cα-RMSD vs native
    rmsdConfident,                 // RMSD in AF-confident regions
    afVerdict:  "af-confirmed" | "af-plausible" | "af-divergent",
    confidentResidueCount,
    category,                      // twin | near-native | variant | conformer | outlier
    ca:         [[x,y,z], …],
    residues:   [ … ]              // same shape as native.residues
  }],

  stats: {
    variantsDiscovered, variantsBeatingNative, energyImprovement,
    energyRange:    { native, best, worstKept },
    searchSpaceExplored: "10^883",
    categoryCounts: { twin, nearNative, variant, conformer, outlier },
    compute:        { searchSeconds, evaluations, evaluationsPerSecond,
                      acceptanceRate, exhaustiveYearsExp, universeAgesExp }
  },

  generatedAt, framework
}

References

1. Anfinsen, C. B. (1973). Principles that govern protein folding. Science, 181(4096), 223-230.
2. Levinthal, C. (1969). How to fold graciously. Mössbauer Spectroscopy in Biological Systems.
3. Dill, K. A., & MacCallum, J. L. (2012). The protein-folding problem, 50 years on. Science, 338(6110), 1042-1046.
4. Jumper, J., Senior, A. W., et al. (2021). Highly accurate protein structure prediction with AlphaFold. Nature, 596(7873), 583-589.
5. Varadi, M., et al. (2022). AlphaFold Protein Structure Database. Nucleic Acids Research, 50(D1), D439-D444.

License

All data is released under CC-BY 4.0. Free to use with attribution. Native structures originate from the AlphaFold Protein Structure Database; please also cite Jumper et al. (2021) and Varadi et al. (2022).