The Ramachandran Plot: Allowed Regions, Outliers, and What They Mean

The 1963 Calculation

In 1963, G. N. Ramachandran, C. Ramakrishnan, and V. Sasisekharan published a systematic calculation of the sterically allowed (φ, ψ) combinations for a polypeptide residue, using hard-sphere van der Waals radii to identify angle combinations that would produce atomic clashes. ^[2] The result was a two-dimensional map of (φ, ψ) space showing that only a small fraction of the possible angle combinations are sterically permissible. What made this calculation immediately and permanently important was the observation that the allowed regions correspond precisely to the regular secondary structures known at the time from model building and emerging crystallographic data: the α-helix, the parallel and antiparallel β-sheets, and the left-handed helix. The Ramachandran plot was not just a steric exclusion diagram. It was a structural prediction that proved correct.

The Three Major Allowed Regions

In the original calculation and all subsequent empirical analyses, the Ramachandran plot for standard L-amino acid residues contains three principal allowed regions. The helical region, centered near (φ = −57°, ψ = −47°), encompasses the right-handed α-helix and the related 3₁₀ and π-helices. The β-sheet region, centered near (φ = −120°, ψ = +135°), encompasses both parallel and antiparallel β-strand conformations and the extended polyproline II helix. The left-handed helical region, centered near (φ = +57°, ψ = +47°), corresponds to the left-handed α-helix and is accessible to standard L-amino acids in principle, though rarely occupied in practice because the side chain clashes that are avoided in right-handed helices are not avoided in the equivalent left-handed geometry. Between these regions lie areas of varying permissibility, including the bridge region between the helical and β-sheet areas that is accessible to glycine and occasionally to alanine.

The total fraction of (φ, ψ) space that is allowed varies by analysis, but hard-sphere calculations place it at approximately 20 to 30 percent of the full 360° × 360° map. Empirical analyses based on high-quality crystal structures refine this further: the core regions, those most densely populated in real structures, account for only about 8 to 10 percent of the total space.

Glycine: The Exception That Expands the Map

Glycine, the only canonical amino acid without a side chain, has a dramatically expanded Ramachandran plot compared to all other residues. The absence of the Cβ atom eliminates a major source of steric constraint: the clashes between the Cβ and the carbonyl oxygen of the preceding residue that restrict φ in all other amino acids do not apply to glycine. As a result, glycine can access virtually all four quadrants of the Ramachandran map, including the left-handed helical region and the bridge region, which are essentially inaccessible to standard L-amino acids. This conformational freedom explains glycine's unique structural roles: in beta turns where an unusual backbone geometry is required, in the third position of collagen's Gly-X-Y repeat where the backbone must pass through the interior of the triple helix, and in loop regions where sharp changes in chain direction are needed.

Proline: The Exception That Restricts the Map

Proline has the most restricted Ramachandran plot of the canonical amino acids. Its pyrrolidine ring constrains φ to a narrow range near −60°, as discussed in Article 2.4. The result is that proline can occupy only a small portion of the helical region and a narrow strip of the β region. It is effectively excluded from the left-handed helical region entirely. Proline-containing peptide bonds also require a separate Ramachandran analysis because they are frequently cis (a significant minority, unlike all other residues), and the cis and trans forms populate different regions of the map. Structure validation software treats proline with a dedicated probability distribution that accounts for this distinct behavior.

From Hard Spheres to Empirical Distributions

The original Ramachandran calculation used hard-sphere van der Waals radii with fixed cutoffs to define allowed and disallowed regions. This approach, while powerful, drew sharp boundaries that did not reflect the actual distribution of angles in real structures, where some nominally disallowed combinations occur due to the flexibility of van der Waals interactions and the stabilizing contributions of hydrogen bonds and electrostatics.

Modern Ramachandran analysis uses empirical probability distributions derived from high-resolution protein crystal structures rather than hard-sphere models. Tools such as MolProbity build their allowed regions from the observed distribution of (φ, ψ) values in structures refined at better than 1.0 Å resolution, using kernel density estimation to define contours that enclose 98 percent and 99.95 percent of the observed data. ^[3] These empirically derived contours are smoother and more nuanced than hard-sphere boundaries, and they incorporate the actual physics of peptide conformational preferences rather than geometric approximations.

What Outliers Mean

A Ramachandran outlier, a residue with a (φ, ψ) combination outside the empirically allowed regions, can mean several distinct things, and distinguishing between them requires judgment rather than automatic flag-and-fix. In a structure determined at low resolution, outliers frequently indicate model-building errors: atoms placed in incorrect positions during refinement. These should be corrected. In a structure determined at high resolution with good experimental data, an outlier indicates either genuine structural strain (the backbone is forced into an unusual conformation by crystal packing, ligand binding, or active site geometry) or a biologically significant unusual conformation maintained for functional reasons. The active sites of enzymes routinely contain residues with strained backbone geometries that are necessary for catalytic function. Non-proline cis peptide bonds, discussed in Article 2.3, represent one common source of legitimate outliers in high-quality structures. A Ramachandran outlier in an otherwise well-refined, high-resolution structure deserves investigation, not reflexive correction.

The Ramachandran Plot as a Universal Diagnostic

The Ramachandran plot remains the single most informative and most widely applied diagnostic for protein and peptide structure quality. Every structure deposited in the Protein Data Bank is evaluated by its Ramachandran statistics, and the percentage of residues in the core region, the allowed region, and in outlier positions is reported as part of the standard validation suite. For peptide structures specifically, where the chain is short and the total number of residues is small, a single outlier represents a larger fraction of the total residue count than in a large protein, making Ramachandran analysis proportionally more important. Any paper reporting a peptide structure should include Ramachandran statistics, and any reader evaluating such a paper should check them.