SASA calculator
v1.2.0Calculate solvent accessible surface area for protein structures
Calculate solvent accessible surface area for protein structures
Upload a PDB or mmCIF file, or fetch an entry from the RCSB PDB, and click Calculate. The calculator rolls a 1.4 Å water probe over every heavy atom with the Shrake-Rupley algorithm and returns solvent accessible surface area (SASA) and relative solvent accessibility (RSA) for each residue, plus chain and whole-structure totals, in one run in your browser.
1UBQ.For ubiquitin (1UBQ), the Summary tab reports one structure:
| Structure | Total SASA (Ų) | Chains | Residues |
|---|---|---|---|
| 1UBQ | 4888.35 | 1 | 76 |
The first rows of the Residues tab show how exposure changes along the chain:
| Chain | Residue | Position | SASA (Ų) | RSA (%) | Exposure | RSA note |
|---|---|---|---|---|---|---|
| A | MET | 1 | 56.59 | Chain terminus | ||
| A | GLN | 2 | 80.01 | 35.6 | Intermediate | |
| A | ILE | 3 | 0.00 | 0.0 | Buried | |
| A | PHE | 4 | 57.76 | 24.1 | Intermediate | |
| A | VAL | 5 | 1.31 | 0.8 | Buried | |
| A | LYS | 6 | 108.08 | 45.8 | Intermediate | |
| A | THR | 7 | 27.84 | 16.2 | Buried | |
| A | LEU | 8 | 134.30 | 66.8 | Exposed |
Ile3 and Val5 sit in the core of the β-sheet, while Leu8 points into solvent. Download any tab as CSV or Excel to sort, filter, or plot the values.
| Input | Details |
|---|---|
| File formats | PDB (.pdb, .ent) and mmCIF (.cif, .mmcif, .pdbx), up to 50 MB per file |
| Database fetch | Any RCSB PDB entry by ID, downloaded as mmCIF so large complexes work too |
| Several files | Each file is calculated separately and gets its own rows, named after the file |
| Predicted models | AlphaFold, ESMFold, Boltz-2 and Chai-1 PDB or mmCIF outputs work as uploaded |
The surface is built from the heavy atoms of protein and nucleic-acid chains:
HETATM records. Selenomethionine (MSE) in crystal structures is the common case; leaving it out would open a hole in the surface and inflate the neighboring residues.HETATM or as ATOM records (as in many molecular dynamics snapshots). The result lists which groups were left out.A pasted sequence or a file without atom coordinates returns an error that says so. SASA needs 3D coordinates; to start from a sequence, predict a structure first.
| Setting | What it does | Options | Default |
|---|---|---|---|
Probe radius (Å) | Radius of the solvent sphere rolled over the surface. 1.4 Å approximates a water molecule. Larger probes skip narrow grooves and report only the most accessible surface. | 1.4 Å (water), 1.8 Å (larger solvent), 2.0 Å | 1.4 Å (water) |
Relative accessibility is only defined for the 1.4 Å probe, because the reference maximum values were calculated with it. With 1.8 or 2.0 Å, the RSA (%) column stays empty and RSA note reads Needs 1.4 Å probe, while absolute SASA is still reported.
Every run fills three tabs from the same calculation:
| Tab | One row per | Columns |
|---|---|---|
| Residues | Residue | Structure, Chain, Residue, Position, Insertion code (shown when present), SASA (Ų), RSA (%), Exposure, RSA note |
| Chains | Chain | Structure, Chain, SASA (Ų), Residues |
| Summary | Structure | Structure, Total SASA (Ų), Chains, Residues |
In the Residues tab, SASA (Ų) is the accessible area of the residue's atoms and RSA (%) divides it by the theoretical maximum for that amino acid. Exposure groups RSA into Buried (below 20%), Intermediate (20 to 50%), and Exposed (above 50%). When RSA is empty, RSA note gives the reason:
Chain terminus for the first and last residue of each chainNo reference value for nucleotides and modified residues without a reference maximumNeeds 1.4 Å probe when a larger probe was usedChain totals are the sum of their residues, and the structure total is the sum of its chains. Totals in the Chains and Summary tabs are calculated from the complex as uploaded, so a chain buried against its partners in hemoglobin (4HHB) reports less area than the same chain on its own.
Notices appear above the results when something in the file changes how the numbers should be read. For 4HHB the notice lists HEM (4), PO4 (2) as left out, which means residues lining the heme pockets are measured as if the heme were absent. For the NMR ensemble 1L2Y it reports that model 1 of 38 was used.
Total SASA tracks the surface of a folded chain, so it grows more slowly than the residue count. These single-chain entries were run with the default settings:
| Entry | Protein | Residues | Total SASA (Ų) | Ų per residue |
|---|---|---|---|---|
1CRN | Crambin | 46 | 2983.99 | 64.9 |
1UBQ | Ubiquitin | 76 | 4888.35 | 64.3 |
1HEL | Hen egg-white lysozyme | 129 | 6750.14 | 52.3 |
1MBN | Sperm whale myoglobin | 153 | 8421.43 | 55.0 |
2LZM | T4 lysozyme | 164 | 8670.43 | 52.9 |
2PTN | Trypsin | 223 | 9337.48 | 41.9 |
Miller et al. (1987) fitted the surface areas of 46 monomeric proteins of 4,000 to 35,000 Da with A = 6.3 M0.73 (A in Ų, M in Da), accurate to 4% on average. Taking M as about 110 Da per residue, the formula lands within about 10% of every row above, from 7% under for trypsin to 10% over for myoglobin. Roughly 65 Ų per residue is normal for a 50 to 80 residue chain, and about 40 to 55 Ų for 130 to 230 residues.
A total well above this range points to an extended or unfolded model, and one well below it to a chain measured inside a complex, where partners cover part of its surface. Two entries of the same protein can differ by a couple of percent: hen lysozyme gives 6750.14 Ų for 1HEL and 6611.40 Ų for 1LYZ. Multi-chain totals add up chains measured inside the complex, so divide by residues only for single chains.
The Shrake-Rupley method, introduced in 1973, places test points on a sphere around each atom and counts how many are not covered by a neighboring atom. Each atom sphere has the atom's van der Waals radius plus the probe radius, so a point on it marks a position the center of a water molecule could reach.
The calculator uses 92 points per atom on a golden-spiral layout and Bondi van der Waals radii: carbon 1.7 Å, nitrogen 1.55 Å, oxygen 1.52 Å, sulfur 1.8 Å, and selenium 1.9 Å. This matches the ShrakeRupley class in Biopython's Bio.PDB.SASA with n_points=92, and residue values agree with it to 0.01 Ų. Compared with a converged calculation using 5000 points, 92 points differ by about 2 Ų per residue on average, which is small next to the differences between radius sets.
The point count is fixed at 92, and the probe radius has three options. For ubiquitin (1UBQ), the 92-point total is 4888.35 Ų against 4870.56 Ų at 5000 points, a 0.4% difference. Per residue, the mean absolute difference is 1.9 Ų and the largest is 7.8 Ų. One test point on a carbon atom stands for 1.31 Ų (4π(1.7 + 1.4)² / 92), so residue values move in steps of that size.
The same coarseness explains some zeros. Four ubiquitin residues read 0.00 at 92 points, while the 5000-point run leaves one. A 0.00 therefore means that no test point escaped, not that the residue is perfectly sealed. Differences of a few Ų between two residues are inside the method's noise.
The probe radius moves totals much less than the sampling does. Ubiquitin gives 4888.35, 4837.27, and 4853.40 Ų at 1.4, 1.8, and 2.0 Š(4870.56, 4852.91, and 4887.79 Ų at 5000 points), within 1% of each other. Use 1.8 or 2.0 Što ask whether something larger than water can reach a site, not to rescale a total; RSA (%) is empty for both.
Accessible surface area depends on the atomic radii and the algorithm. FreeSASA uses the Lee-Richards slicing algorithm with ProtOr radii by default, and DSSP uses its own atomic radii, so totals from different programs usually differ by a few percent. Compare structures with one method and one probe radius, and report both when you publish SASA values.
SASA is also different from the solvent-excluded (molecular or Connolly) surface. SASA traces the center of the probe, so it sits one probe radius outside the atoms; the solvent-excluded surface traces the probe's contact with the atoms and encloses a smaller area.
Relative solvent accessibility compares a residue's SASA with the largest accessible area that amino acid can have:
Maximum values come from the theoretical Gly-X-Gly scale of Tien et al. (2013), which ranges from 104 Ų for glycine to 285 Ų for tryptophan. Selenomethionine uses the methionine value, and force-field residue names from molecular dynamics (such as HIE, HID, and HIP for histidine, CYX for cystine, ASH, GLH, and LYN) use their parent amino acid.
Chain termini have no RSA, since their free amino or carboxyl group has no counterpart in the Gly-X-Gly reference. Values above 100% stay visible rather than being capped; they usually point to an extended loop or a residue with missing neighbors.
The Exposure column uses below 20% for Buried and above 50% for Exposed, with Intermediate in between. These boundaries are a three-state convention. Tien et al. supply the maximum values, not a cutoff, and published studies disagree on where buried begins. The cutoff changes the answer: ubiquitin (1UBQ) has 74 residues with an RSA value, and the count of buried ones depends on the threshold.
| RSA cutoff | Residues in 1UBQ | Share |
|---|---|---|
| Below 5% | 14 | 19% |
| Below 10% | 16 | 22% |
| Below 20% | 24 | 32% |
| Below 25% | 28 | 38% |
| Above 50% | 16 | 22% |
Miller et al. (1987) called a residue buried when less than 5% of its surface was accessible, and found that 15% of residues in small proteins and 32% in larger ones qualified. Levy (2010) tested interior against surface residues in E. coli, yeast, and human structures and found that 25% RSA separates them best, which is why 25% is a common choice. Whichever cutoff you use, state it next to the result.
Tien et al. calculated the Gly-X-Gly maxima with DSSP and describe them as strictly valid for DSSP areas. This calculator uses Shrake-Rupley areas with Bondi radii, so RSA here can differ slightly from DSSP accessibility divided by the same maxima. The new maxima are up to 20% higher than the older Miller and Rose tables, which is why values above 100% are rare now; percentages computed against an older table are not comparable.
The atoms in the calculation matter more than the settings. In sperm whale myoglobin (1MBN) the heme is left out; with it in place, the protein's SASA falls from 8421.43 to 7926.0 Ų (495 Ų, 6%). Biotin in streptavidin (1STP) removes 178 Ų (3%). Residues in a binding pocket therefore read as more exposed than they are in the liganded complex.
Crystal waters are excluded because they would cover the protein. With its crystallographic waters kept, ubiquitin falls from 4888.35 to 3379.1 Ų, and 1HEL lysozyme from 6750.14 to 3375.7 Ų. These figures come from adding the groups back outside the calculator, which has no option for it.
Every polymer chain in the file is included, and there is no chain selector. To measure one chain alone, upload a file that contains only that chain. Hemoglobin chain A reports 6198.70 Ų in the 4HHB tetramer and 7904.18 Ų alone.
Two points about chain handling are worth knowing. A modified residue stays in when a nitrogen or carbon of it is peptide-bonded to a neighboring chain residue, so selenomethionine and the GFP chromophore CRO in 1EMA are kept even though CRO has no atoms named N and C. Chain terminus marks the first and last residue present in each chain, so residues beside an internal gap in the model are not flagged and may read high.
Surface exposure tells you which residues can meet solvent, binding partners, or modifying enzymes. Before a mutation, it confirms that a target residue is on the surface, or that a core position is buried before a stabilizing substitution. Exposed hydrophobic residues form sticky patches, which is why aggregation propensity scores combine exposure with residue chemistry.
The area buried at an interface equals the SASA of the two chains calculated alone minus the SASA of the complex. Upload the complex and each isolated chain as separate files, then subtract the totals in the Summary tab.
In hemoglobin (4HHB), chain A alone has 7904.18 Ų, chain B alone 8304.78 Ų, and the A and B pair together 14592.05 Ų, so their interface buries 1616.9 Ų.
For barnase (chain A) and barstar (chain D) from 1BRS, 5927.21 + 4925.00 − 9300.36 gives 1551.8 Ų. This sum counts both partners; halve it for the area lost by each side, and say which you report. To turn a complex into an affinity estimate, use PRODIGY.
In AlphaFold or ESMFold models, low-confidence regions are often extended and report high RSA that may not reflect the native protein.
Missing atoms lower the calculated surface. Crystal structures often lack side chains for flexible lysine and glutamate residues; rebuilding them with PDBFixer before calculating gives more complete values.
A residue labeled Exposed is not necessarily free in the cell. The label compares areas in a structure stripped of its ligands, partners, and waters, so an active-site residue can read Exposed while the real substrate covers it.
RSA moves more for small residues. A 2 Ų change, about the sampling noise, shifts glycine (maximum 104 Ų) by 1.9 percentage points and tryptophan (285 Ų) by 0.7. Near a cutoff, a residue at 19.5% and one at 20.5% are the same residue.
Total SASA is a surface, not a size. Two proteins with the same total can differ in compactness, so pair it with the radius of gyration. In a predicted model, disordered tails stick into solvent and report high RSA whatever their function; check pLDDT before interpreting them.
| Goal | Tool |
|---|---|
| Surface area and residue exposure | SASA calculator |
| Secondary structure with per-residue accessibility | DSSP |
| Aggregation-prone surface patches | Aggrescan3D |
| Binding pockets and cavities | fpocket |
| Backbone geometry | Ramachandran plot |
| Overall compactness | Radius of gyration |
| Differences between two structures | RMSD calculator |
| SASA over a simulation | MD trajectory analysis |
Use 1.4 Å, the approximate radius of a water molecule. It is the standard in the literature and the only probe that supports relative solvent accessibility.
Residues below 20 to 25% RSA are usually called buried, and residues above 50% exposed. The calculator marks residues below 20% as Buried and above 50% as Exposed.
No. Ligands, cofactors, ions, and solvent are left out, and the result lists the groups that were removed. Residues in a binding site are therefore measured as if the ligand were absent.
Yes. The reference maximum comes from a short Gly-X-Gly peptide, so a residue in a fully extended loop or next to missing atoms can have slightly more accessible area. Such values are reported as calculated.
Upload the model's PDB or mmCIF file from the AlphaFold Database or from a prediction run. The calculator reads it like any other structure; check low-confidence regions before interpreting their exposure.
Export the frames you need as PDB files and upload them together; each frame gets its own rows. For whole trajectories, use MD trajectory analysis, which calculates SASA frame by frame.
Assign helices, sheets, turns, bends, and PPII structure from atomic coordinates.
Validate protein structures with clashscore, Ramachandran, rotamer, and geometry checks.
Generate structural summary reports and diagrams for a single protein PDB structure.
Pairwise RMSD matrices for PDB structure ensembles
Calculate the radius of gyration (Rg) for protein structures from PDB files.
Calculate RMSD between protein structures with automatic alignment
Predict metal and water binding sites in proteins with 3D CNN models.
Score every protein interface with ESM-2 and a graph neural network
Assess docking quality between model and native structures
Score interprotein interactions in AlphaFold and Boltz predictions