DSSP

What is DSSP?

DSSP (Dictionary of Secondary Structure of Proteins) is the gold standard algorithm for assigning protein secondary structure from atomic coordinates. Developed by Kabsch and Sander in 1983, DSSP analyzes hydrogen bonding patterns in protein backbones to classify each residue as helix, sheet, turn, bend, or loop.

Unlike methods that rely on annotated HELIX and SHEET records in PDB files, DSSP calculates secondary structure directly from atomic positions. This makes it ideal for analyzing predicted structures from AlphaFold, ESMFold, or Boltz-2, which typically lack these annotations.

DSSP remains the most widely used secondary structure assignment tool in structural biology, serving as the reference standard against which other methods are validated.

How does DSSP work?

Hydrogen bond detection

DSSP identifies backbone hydrogen bonds using the Kabsch-Sander electrostatic model. The algorithm assumes partial charges of $q_1 = 0.42e$ on the carbonyl oxygen and $q_2 = 0.20e$ on the amide hydrogen.

The hydrogen bond energy between a donor N-H and acceptor C=O is calculated as:

$$E = q_1 q_2 \left( \frac{1}{r_{ON}} + \frac{1}{r_{CH}} - \frac{1}{r_{OH}} - \frac{1}{r_{CN}} \right) \times 332 \text{ kcal/mol}$$

where $r_{ON}$, $r_{CH}$, $r_{OH}$, and $r_{CN}$ are distances (in Ångströms) between the respective atom pairs. A hydrogen bond exists when $E < -0.5$ kcal/mol.

Secondary structure codes

DSSP assigns eight structure types based on hydrogen bonding patterns:

Assignment algorithm

Helices form when consecutive n-turns overlap. An n-turn exists when residue i+n donates a hydrogen bond to residue i. Two consecutive 4-turns create an α-helix (H), two consecutive 3-turns create a 3₁₀-helix (G), and two consecutive 5-turns create a π-helix (I).

β-strands are identified through ladder patterns where residues form parallel or antiparallel hydrogen-bonded bridges. A single bridge is classified as B, while consecutive bridges form extended strands (E).

Bends are assigned when the angle between Cα positions at i-2, i, and i+2 exceeds 70°, indicating a sharp change in backbone direction.

Inputs & settings

Structure input

Upload PDB files or fetch structures using 4-letter PDB IDs. DSSP requires complete backbone atoms (N, CA, C, O) for each residue. Missing backbone atoms will cause those residues to be skipped.

Output level

Summary — Provides per-chain percentages of helix, sheet, and loop content. This condensed view is useful for comparing overall structural composition across multiple proteins.

Per residue — Shows DSSP codes for every amino acid along with hydrogen bond partners. Use this detailed output to identify specific structural elements or analyze local folding patterns.

Understanding the results

Summary output

The summary table shows structural composition for each chain:

• Helix (%) — Combined percentage of all helix types (H + G + I)
• Sheet (%) — Combined percentage of strands and bridges (E + B)
• Loop (%) — Combined percentage of turns, bends, and coils (T + S + space)

Per-residue output

Each row represents one amino acid with its assigned structure code and hydrogen bond partners:

• DSSP Code — Single-letter code (H, G, I, E, B, T, S, or space)
• N→O Partners — Residue numbers where this residue's N-H donates hydrogen bonds
• O←N Partners — Residue numbers where this residue's C=O accepts hydrogen bonds

Hydrogen bond partners help identify interaction patterns. For example, residues in an α-helix show i→i+4 bonding patterns.

Interpreting DSSP codes

α-helices (H) are the most common secondary structure, typically comprising 30-35% of globular proteins. They form stable, right-handed helical structures with 3.6 residues per turn.

β-strands (E) usually represent 20-25% of protein structure. They combine with other strands to form β-sheets, which can be parallel, antiparallel, or mixed.

Loops and coils (space) connect structured elements and often contain functionally important binding sites or catalytic residues. They typically make up 15-25% of the structure.

3₁₀-helices (G) are shorter and tighter than α-helices, often appearing at helix ends or in tight turns. π-helices (I) are rare, comprising less than 1% of residues.

Limitations

DSSP requires complete backbone atoms (N, CA, C, O) for accurate assignment. Structures with missing atoms will have incomplete results. Use PDB Fixer to add missing atoms before running DSSP.

The algorithm is sensitive to coordinate precision. Low-resolution structures or poorly refined models may produce unreliable assignments, particularly for borderline cases like short helices or isolated bridges.

DSSP cannot distinguish between static and dynamic secondary structure. Flexible regions that transiently adopt helical or extended conformations are assigned based on their conformation in the input structure.

Related tools

• STRIDE — Alternative secondary structure assignment using hydrogen bonds and backbone torsion angles
• Ramachandran Plot — Validate backbone geometry through φ/ψ angle distribution
• Radius of Gyration — Measure overall protein compactness
• RMSD Calculator — Compare structural similarity between proteins
• PDB Viewer — Visualize protein structures with secondary structure coloring