Ramachandran Plot

A Ramachandran plot (also termed a Ramachandran diagram or [φ, ψ] plot) is a fundamental two-dimensional conformational coordinate map in structural biology and protein biophysics that displays the sterically permitted values of the backbone dihedral torsion angles ψ (psi) on the vertical axis (−180° to +180°) versus φ (phi) on the horizontal axis (−180° to +180°) for amino acid residues in a polypeptide chain. Developed in 1963 at the University of Madras by the Indian biophysicist G. N. Ramachandran together with C. Ramakrishnan and V. Sasisekharan, the plot demonstrated mathematically—using hard-sphere van der Waals contact radii before a single atomic-resolution crystal structure of a globular protein had been fully refined—that more than 75% of the (φ, ψ) conformational space of a polypeptide backbone is sterically forbidden due to short-range atomic collisions.

Because the repeating amide peptide bond is locked into a rigid planar conformation, the three-dimensional trajectory of any protein backbone is determined almost entirely by the sequence of its (φ_i, ψ_i) torsion pairs. In modern structural biology, the Ramachandran plot serves both as the primary classification map for secondary structure elements—such as right-handed α-helices, parallel and antiparallel β-sheets, 3₁₀-helices, and polyproline II helices—and as the gold-standard geometric validation tool in X-ray crystallography, cryogenic electron microscopy (cryo-EM), NMR spectroscopy, and deep-learning structure prediction (such as AlphaFold).

Fold Polypeptides in 3D: Explore how backbone torsion angles (φ, ψ) and side-chain interactions drive secondary and tertiary structure in our interactive FoldLab simulation!

Planarity of the Peptide Bond (ω) and Rotational Freedom of φ and ψ

The covalent main chain of a polypeptide consists of a repeating three-bond unit for each amino acid residue i: –N_i–C_α,i–C′_i–N_i+1–, where N is the amide nitrogen, C_α is the chiral alpha-carbon bearing the side chain R, and C′ is the carbonyl carbon. The conformation of each residue i is specified by three internal dihedral torsion angles defined by four sequential backbone atoms:

1. Omega (ω), Defined by C_α,i–C′_i–N_i+1–C_α,i+1 (Rotation Around the C′–N Peptide Bond): As shown by Linus Pauling and Robert Corey in 1951 and explained by Lewis structure resonance delocalization of the nitrogen lone pair into the carbonyl π* orbital, the peptide C′–N bond possesses ~40% double-bond character (bond length 1.33 Å vs 1.47 Å for a single C–N bond) with a rotational barrier of ~84 kJ mol⁻¹ (~20 kcal mol⁻¹). Consequently, rotation around ω is restricted: >99.7% of non-proline peptide bonds adopt the planar trans conformation (ω ≈ 180°), which keeps sequential C_α atoms at maximal separation (3.80 Å), while ~0.03% of non-proline bonds and ~5–6% of Xaa–Pro bonds adopt the planar cis conformation (ω ≈ 0°) (C_α–C_α distance = 2.95 Å).
2. Phi (φ), Defined by C′_i−1–N_i–C_α,i–C′_i (Rotation Around the N–C_α Single Bond): Because the N–C_α bond (1.46 Å) is a pure σ single bond, rotation around φ is intrinsically free except where blocked by steric clashes between flanking carbonyl groups and the C_β side chain. By IUPAC–IUB convention, φ = 180° when the two carbonyl carbons (C′_i−1 and C′_i) are trans (antiperiplanar), and positive angles (+0° to +180°) correspond to clockwise rotation of the rear group relative to the front group when viewed down the N–C_α bond.
3. Psi (ψ), Defined by N_i–C_α,i–C′_i–N_i+1 (Rotation Around the C_α–C′ Single Bond): Similarly, the C_α–C′ bond (1.52 Å) is a pure σ single bond whose rotation angle ψ equals 180° when the two amide nitrogens (N_i and N_i+1) are trans.

With ω fixed near 180°, the entire spatial conformation of residue i reduces to a single point (φ_i, ψ_i) on the two-dimensional Ramachandran plane.

Hard-Sphere Van der Waals Clash Limits

To construct the original 1963 diagram, Ramachandran and colleagues modeled an L-alanyl dipeptide mimic (N-acetyl-L-alanine-methylamide, containing a C_βH₃ methyl side chain) as a collection of rigid spheres assigned two sets of minimum non-bonded contact distances: a "normally allowed" van der Waals contact limit (e.g., C···C ≥ 3.20 Å, O···O ≥ 2.80 Å, N···O ≥ 2.70 Å) and a slightly shorter "outer limit" (e.g., C···C ≥ 3.00 Å, O···O ≥ 2.70 Å). Scanning φ and ψ across all 360° × 360° combinations revealed that most of the map is forbidden by three specific short-range steric collisions:

• The φ ≈ 0° Vertical Barrier: Rotating φ near 0° forces the carbonyl carbon and oxygen of the preceding residue (C′_i−1=O_i−1) directly into the carbonyl group of the current residue (C′_i=O_i).
• The ψ ≈ 0° Horizontal Barrier: Rotating ψ near 0° forces the amide NH of the current residue (N_i–H_i) to collide with the amide NH of the following residue (N_i+1–H_i+1).
• The Right-Half (φ > 0°) Chiral C_β Steric Exclusion: For any L-amino acid possessing a C_β carbon atom (all 19 standard chiral amino acids except glycine), positive values of φ (the right half of the plot, φ > 0°) bring the C_β side chain into severe steric collision with the preceding peptide carbonyl oxygen (O_i−1) or following amide group, eliminating nearly all of the right half of the plot except for a small island around the left-handed α_L-helix.

Consequently, non-glycine L-amino acids are confined to three distinct allowed basins (which occupy ~15–20% of the total area under core criteria):

1. The Right-Handed α-Helical Basin (Lower-Left Quadrant, φ ≈ −60°, ψ ≈ −45°): Contains the canonical right-handed α-helix (3.6 residues per turn, i + 4 → i hydrogen bond, φ = −57°, ψ = −47°), the tighter 3₁₀-helix (3.0 residues per turn, i + 3 → i hydrogen bond, φ = −49°, ψ = −26°), and the wider π-helix (4.4 residues per turn, i + 5 → i hydrogen bond, φ = −57°, ψ = −70°). Note that in early hard-sphere calculations, the α-helix and β-sheet regions were separated by a narrow "bridge" near ψ ≈ 0° caused by relaxing N_i···N_i+1 dipolar interactions.
2. The Extended β-Strand and Polyproline II Basin (Upper-Left Quadrant, φ ≈ −60° to −150°, ψ ≈ +110° to +170°): This broad upper-left region houses antiparallel β-sheets (φ = −139°, ψ = +135°), parallel β-sheets (φ = −119°, ψ = +113°), twisted β-strands, and the left-handed polyproline II (PPII) helix (φ = −75°, ψ = +145°, 3.0 residues per turn with no intrachain H-bonds), which forms the triple-helical strands of collagen and dominates the conformational ensemble of intrinsically disordered proteins (IDPs).
3. The Left-Handed α_L-Helical Region (Upper-Right Quadrant, φ ≈ +57°, ψ ≈ +47°): Because L-amino acid C_β atoms clash slightly with the backbone at positive φ, continuous left-handed helices of L-amino acids are rare, but single L-residues frequently adopt the α_L conformation at the corner of tight Type I′ and Type II β-turns and in functional active site loops.

Glycine: Achiral Conformational Freedom Across All Four Quadrants

Among the 20 standard proteinogenic amino acids, glycine (Gly, R = H) is unique because its C_α atom is bonded to two hydrogen atoms rather than a bulky C_β carbon (van der Waals radius of H ≈ 1.20 Å vs CH₃ group ≈ 2.00 Å). Because glycine is achiral and lacks a C_β substituent:

• Its Ramachandran plot is inversion-symmetric through the origin: every conformation allowed at (φ, ψ) is equally allowed at (−φ, −ψ).
• Well over 45–50% of the total Ramachandran area is sterically accessible to glycine across all four quadrants. Consequently, glycine is evolutionarily conserved in structural niches inaccessible to any other L-amino acid, including the tight central axes of the collagen triple helix (every third residue in the (Gly–Pro–Hyp)_n repeat), sharp Type II β-turns (where the i + 2 residue requires φ ≈ +80°, ψ ≈ 0°), and flexible hinge regions of enzymes. Conversely, in unfolded polypeptide chains, glycine possesses the highest conformational entropy (ΔS_conf), making it thermodynamically unfavorable inside rigid α-helices.

Proline and Pre-Proline: Cyclic Restriction

At the opposite extreme of conformational rigidity sits proline (Pro), an imino acid whose three-carbon side chain cyclizes back onto its own backbone amide nitrogen to form a five-membered pyrrolidine ring:

• Because the N–C_α bond is embedded inside the five-membered ring, the φ torsion angle of proline is covalently locked to φ = −60° ± 15°, restricting proline to two narrow islands at (φ ≈ −60°, ψ ≈ −35° in α-helices) and (φ ≈ −75°, ψ ≈ +145° in polyproline II / collagen helices).
• Furthermore, because the backbone nitrogen of proline lacks an amide proton (–NH–) and carries a bulky C_δH₂ methylene group, proline cannot donate an i + 4 → i hydrogen bond inside an α-helix and sterically restricts the preceding residue (pre-proline, i − 1), excluding pre-proline residues from the α_R and α_L regions when ψ_i−1 lies near −50° to +30°.

In contemporary structural biology software—such as PROCHECK (Roman Laskowski, 1993) and MolProbity (Jane and David Richardson)—the classic hard-sphere Ramachandran plot has been replaced by empirical Bayesian probability contours derived from millions of ultrahigh-resolution (<1.2 Å), low-B-factor crystal structures in the Protein Data Bank (PDB). MolProbity classifies residues into six distinct Ramachandran distributions (General non-Gly/non-Pro/non-Ile/Val, Ile/Val β-branched, Glycine, trans-Proline, cis-Proline, and Pre-Proline) and divides the plot into:

• Favored Regions (inner 98% contour): High-resolution, well-refined protein structures contain >98.0% of all residues inside the favored contours.
• Allowed Regions (99.8% outer contour): Typically <2.0% of residues lie in the marginal allowed belt.
• Ramachandran Outliers (<0.05% to 0.2% of residues): Because forcing a peptide backbone into a forbidden (φ, ψ) angle incurs >15–25 kJ mol⁻¹ of van der Waals strain, a cluster of Ramachandran outliers almost always diagnoses local model-building errors or poor electron-density fitting in X-ray or cryo-EM models. When a true Ramachandran outlier does exist in an atomic-resolution crystal structure, it is held in place by multiple compensating hydrogen bonds and invariably marks a strained functional residue inside an enzymatic active site or ligand-binding pocket.

What would the Ramachandran plot of a protein made entirely of D-amino acids look like?

Replacing all L-amino acids in a protein with their non-superimposable mirror-image D-amino acid enantiomers (as constructed by total chemical synthesis of mirror-image enzymes such as D-HIV-1 protease) inverts the chirality of every C_α stereocenter. Geometrically, reflection through a mirror plane transforms every dihedral torsion angle (φ, ψ) into (−φ, −ψ)—an exact point inversion (180° rotation) of the Ramachandran plot through the origin (0°, 0°). Thus, for a D-amino acid polypeptide, the left-handed α-helix becomes the major lower-energy helical basin in the upper-right quadrant (φ = +57°, ψ = +47°), the β-sheet basin moves to the lower-right quadrant (φ ≈ +119° to +139°, ψ ≈ −113° to −135°), and the rare α-turn island moves to the lower-left quadrant (−57°, −47°).

Why do β-branched amino acids (isoleucine, valine, and threonine) have a more restricted Ramachandran plot?

Unlike alanine (–CH₃) or leucine (–CH₂–CH(CH₃)₂), which have only one non-hydrogen substituent attached to their C_β carbon, the three β-branched amino acids—valine (Val), isoleucine (Ile), and threonine (Thr)—bear two heavy-atom substituents directly on C_β. Regardless of how the side-chain χ₁ torsion angle rotates, at least one of those two C_γ branches is forced into close proximity with the local polypeptide backbone. This added steric bulk shrinks the allowed (φ, ψ) space inside the α-helical basin (causing Val and Ile to destabilize α-helices relative to alanine) and shifts their favored distribution toward the more extended β-sheet basin in the upper-left quadrant, where the backbone atoms are splayed apart.

Why can rare Ramachandran outliers occur legitimately inside enzyme active sites?

While >99.8% of residues in well-refined protein structures obey Ramachandran steric limits, roughly 0.05% of residues in ultrahigh-resolution (<1.0 Å) crystal structures exhibit genuine, unambiguous electron density in a sterically forbidden region (such as the "epsilon" region near φ ≈ +70°, ψ ≈ −120°). During protein folding, the huge favorable free energy gained from burying the hydrophobic core and forming hundreds of tertiary hydrogen bonds (~400–800 kJ mol⁻¹ total) allows the global protein scaffold to pay a localized 15–25 kJ mol⁻¹ steric strain penalty at a single catalytic residue. This pre-strained backbone conformation (an organic analog of the entatic state) precisely positions a backbone NH group for an oxyanion hole or aligns a catalytic side chain for transition-state stabilization.

1. Ramachandran, G. N., Ramakrishnan, C., & Sasisekharan, V. (1963). Stereochemistry of polypeptide chain configurations. Journal of Molecular Biology, 7(1), 95–99. https://doi.org/10.1016/S0022-2836(63)80023-680023-6)
2. Lovell, S. C., Davis, I. W., Arendall, W. B., de Bakker, P. I. W., Word, J. M., Prisant, M. G., Richardson, J. S., & Richardson, D. C. (2003). Structure validation by Cα geometry: φ, ψ and Cβ deviation. Proteins: Structure, Function, and Bioinformatics, 50(3), 437–450. https://doi.org/10.1002/prot.10286
3. Laskowski, R. A., MacArthur, M. W., Moss, D. S., & Thornton, J. M. (1993). PROCHECK: A program to check the stereochemical quality of protein structures. Journal of Applied Crystallography, 26(2), 283–291. https://doi.org/10.1107/S0021889892009944
4. Ho, B. K., Thomas, A., & Brasseur, R. (2003). Revisiting the Ramachandran plot: Hard-sphere repulsion, electrostatics, and H-bonding in the α-helix. Protein Science, 12(11), 2508–2522. https://doi.org/10.1110/ps.03235203