zsasa 0.2.0

Fast SASA calculation using Zig

zsasa Python Bindings

Python bindings for the zsasa library - a high-performance implementation of Solvent Accessible Surface Area (SASA) calculation.

Features

• NumPy integration: Pass coordinates and radii as NumPy arrays
• Two algorithms: Shrake-Rupley (point-based) and Lee-Richards (slice-based)
• Multi-threading: Automatic parallelization across CPU cores
• Atom classification: NACCESS, PROTOR, and OONS classifiers with polar/apolar assignment
• RSA calculation: Relative Solvent Accessibility using standard reference values
• Per-residue aggregation: Aggregate atom-level SASA to residue-level with RSA
• Library integrations: Built-in support for gemmi, BioPython, and Biotite
• Type-safe: Full type hints and runtime validation

Installation

From Source (Development)

# 1. Build the Zig shared library
cd /path/to/zsasa
zig build -Doptimize=ReleaseFast

# 2. Install Python package
cd python
pip install -e .

Optional Dependencies

# For gemmi integration
pip install zsasa[gemmi]

# For BioPython integration
pip install zsasa[biopython]

# For Biotite integration (also works with AtomWorks)
pip install zsasa[biotite]

# All integrations
pip install zsasa[all]

Library Location

The Python bindings look for libzsasa.dylib (macOS), libzsasa.so (Linux), or zsasa.dll (Windows) in these locations:

1. ZSASA_LIB environment variable (if set)
2. ../zig-out/lib/ (relative to package)
3. /usr/local/lib/
4. /usr/lib/
5. Current working directory

Quick Start

Basic SASA Calculation

import numpy as np
from zsasa import calculate_sasa, get_version

print(f"Library version: {get_version()}")

# Define atom coordinates (N, 3) and radii (N,)
coords = np.array([
    [0.0, 0.0, 0.0],
    [3.0, 0.0, 0.0],
    [6.0, 0.0, 0.0],
])
radii = np.array([1.5, 1.5, 1.5])

# Calculate SASA
result = calculate_sasa(coords, radii)
print(f"Total SASA: {result.total_area:.2f} Ų")
print(f"Per-atom areas: {result.atom_areas}")

Using Library Integrations (Recommended)

The easiest way to calculate SASA from structure files:

# With gemmi (fast mmCIF/PDB parser)
from zsasa.integrations.gemmi import calculate_sasa_from_structure
result = calculate_sasa_from_structure("protein.cif")

# With BioPython
from zsasa.integrations.biopython import calculate_sasa_from_structure
result = calculate_sasa_from_structure("protein.pdb")

# With Biotite (also works with AtomWorks)
from zsasa.integrations.biotite import calculate_sasa_from_structure
result = calculate_sasa_from_structure("protein.pdb")

print(f"Total: {result.total_area:.1f} Ų")
print(f"Polar: {result.polar_area:.1f} Ų")
print(f"Apolar: {result.apolar_area:.1f} Ų")

Per-Residue Analysis with RSA

from zsasa.integrations.gemmi import calculate_sasa_from_structure
from zsasa.analysis import aggregate_from_result

# Calculate SASA
result = calculate_sasa_from_structure("protein.cif")

# Aggregate to per-residue
residues = aggregate_from_result(result)

for res in residues:
    rsa_str = f"{res.rsa:.1%}" if res.rsa is not None else "N/A"
    print(f"{res.chain_id}:{res.residue_name}{res.residue_id}: "
          f"{res.total_area:.1f} Ų (RSA: {rsa_str})")

API Reference

Core Functions

calculate_sasa(coords, radii, *, algorithm="sr", n_points=100, n_slices=20, probe_radius=1.4, n_threads=0)

Calculate Solvent Accessible Surface Area.

Parameters:

Parameter	Type	Default	Description
coords	NDArray[float64]	required	Atom coordinates as (N, 3) array
radii	NDArray[float64]	required	Atom radii as (N,) array
algorithm	"sr" | "lr"	"sr"	Shrake-Rupley or Lee-Richards
n_points	int	100	Test points per atom (SR only)
n_slices	int	20	Slices per atom (LR only)
probe_radius	float	1.4	Water probe radius in Å
n_threads	int	0	Number of threads (0 = auto-detect)

Returns: SasaResult

SasaResult

Attribute	Type	Description
total_area	float	Total SASA in Ų
atom_areas	NDArray[float64]	Per-atom SASA values

classify_atoms(residue_names, atom_names, classifier=NACCESS)

Classify atoms and get radii.

Parameters:

Parameter	Type	Default	Description
residue_names	list[str]	required	Residue names (e.g., ["ALA", "GLY"])
atom_names	list[str]	required	Atom names (e.g., ["CA", "N"])
classifier	ClassifierType	NACCESS	NACCESS, PROTOR, or OONS
include_classes	bool	True	Whether to include atom classes

Returns: ClassificationResult with radii and classes arrays

get_version()

Returns the library version string (e.g., "0.1.0").

calculate_rsa(sasa, residue_name)

Calculate Relative Solvent Accessibility.

Parameters:

Parameter	Type	Description
sasa	float	SASA value in Ų
residue_name	str	3-letter residue code

Returns: float | None (None for non-standard amino acids)

Utility Functions

These functions are available for advanced use cases:

Function	Description
get_radius(residue, atom, classifier)	Get radius for a specific atom
get_atom_class(residue, atom, classifier)	Get polarity class for an atom
guess_radius(element)	Guess radius from element symbol
guess_radius_from_atom_name(atom_name)	Guess radius from PDB atom name
get_max_sasa(residue_name)	Get max SASA reference value for RSA
calculate_rsa_batch(sasa_values, residue_names)	Batch RSA calculation

Analysis Functions

aggregate_from_result(result)

Aggregate per-atom SASA to per-residue.

Parameters:

Parameter	Type	Description
result	SasaResultWithAtoms	Result from integration module

Returns: list[ResidueResult]

ResidueResult

Attribute	Type	Description
chain_id	str	Chain identifier
residue_id	int	Residue sequence number
residue_name	str	3-letter residue code
total_area	float	Total SASA in Ų
polar_area	float	Polar SASA in Ų
apolar_area	float	Apolar SASA in Ų
rsa	float | None	Relative Solvent Accessibility
n_atoms	int	Number of atoms

Integration Modules

All integration modules provide the same API:

• extract_atoms_from_model(model) - Extract atom data
• calculate_sasa_from_model(model) - Calculate SASA from model object
• calculate_sasa_from_structure(source) - Calculate SASA from file or structure

SasaResultWithAtoms

Extended result with atom metadata:

Attribute	Type	Description
total_area	float	Total SASA in Ų
atom_areas	NDArray[float64]	Per-atom SASA values
atom_classes	NDArray[int32]	Per-atom polarity classes
atom_data	AtomData	Atom metadata
polar_area	float	Total polar SASA in Ų
apolar_area	float	Total apolar SASA in Ų

Enums

ClassifierType

Value	Description
NACCESS	NACCESS classifier (default)
PROTOR	ProtOr classifier
OONS	OONS classifier

AtomClass

Value	Description
POLAR	Polar atom (N, O, etc.)
APOLAR	Apolar atom (C, S, etc.)
UNKNOWN	Unknown classification

Examples

Algorithm Comparison

import numpy as np
from zsasa import calculate_sasa

coords = np.array([[0.0, 0.0, 0.0], [3.0, 0.0, 0.0]])
radii = np.array([1.5, 1.5])

# Shrake-Rupley (default, faster)
result_sr = calculate_sasa(coords, radii, algorithm="sr", n_points=100)
print(f"SR: {result_sr.total_area:.2f} Ų")

# Lee-Richards (more precise)
result_lr = calculate_sasa(coords, radii, algorithm="lr", n_slices=20)
print(f"LR: {result_lr.total_area:.2f} Ų")

Multi-threading

# Auto-detect CPU cores (default)
result = calculate_sasa(coords, radii, n_threads=0)

# Use specific number of threads
result = calculate_sasa(coords, radii, n_threads=4)

# Single-threaded
result = calculate_sasa(coords, radii, n_threads=1)

Atom Classification

from zsasa import classify_atoms, ClassifierType, AtomClass

residue_names = ["ALA", "ALA", "ALA"]
atom_names = ["N", "CA", "O"]

# With default classifier (NACCESS)
result = classify_atoms(residue_names, atom_names)

# Or with explicit classifier
result = classify_atoms(residue_names, atom_names, ClassifierType.PROTOR)

print(f"Radii: {result.radii}")
print(f"Classes: {result.classes}")

# Count polar/apolar
polar_count = sum(1 for c in result.classes if c == AtomClass.POLAR)
apolar_count = sum(1 for c in result.classes if c == AtomClass.APOLAR)
print(f"Polar: {polar_count}, Apolar: {apolar_count}")

Finding Buried Residues

from zsasa.integrations.gemmi import calculate_sasa_from_structure
from zsasa.analysis import aggregate_from_result

result = calculate_sasa_from_structure("protein.cif")
residues = aggregate_from_result(result)

# Find buried residues (RSA < 20%)
buried = [r for r in residues if r.rsa is not None and r.rsa < 0.2]

print(f"Buried residues ({len(buried)}):")
for r in buried:
    print(f"  {r.chain_id}:{r.residue_name}{r.residue_id} - RSA: {r.rsa:.1%}")

Working with AtomWorks

# AtomWorks is built on Biotite, so use the biotite integration
from atomworks.io.utils.io_utils import load_any
from zsasa.integrations.biotite import calculate_sasa_from_atom_array
from zsasa.analysis import aggregate_from_result

# Load with AtomWorks
atom_array = load_any("protein.cif.gz")

# Calculate SASA with zsasa
result = calculate_sasa_from_atom_array(atom_array)
residues = aggregate_from_result(result)

print(f"Total SASA: {result.total_area:.1f} Ų")

Performance

Library-to-library comparison (Python bindings vs FreeSASA Python):

Structure	Atoms	Zig SR	FS SR	SR Speedup	Zig LR	FS LR	LR Speedup
1CRN	327	0.5ms	0.7ms	1.4x	1.5ms	4.4ms	2.9x
1UBQ	602	0.6ms	1.3ms	2.2x	2.0ms	8.4ms	4.2x
1A0Q	3,183	2.4ms	7.6ms	3.2x	8.9ms	48ms	5.5x
3HHB	4,384	3.4ms	11ms	3.2x	12ms	69ms	5.5x
1AON	58,674	44ms	163ms	3.7x	171ms	931ms	5.5x

• SR algorithm: Zig is 1.4-3.7x faster (speedup increases with size)
• LR algorithm: Zig is 2.9-5.5x faster
• Accuracy: Results match FreeSASA (< 0.01% difference)

Run benchmark: ./benchmarks/scripts/run.py --tool zig --algorithm sr

Development

Running Tests

cd python
uv run --with pytest pytest tests/ -v

Type Checking

cd python
uv run --with ty ty check zsasa/

Linting

ruff format .
ruff check --fix .

References

RSA Reference Values

• MaxSASA: Tien, M. Z.; Meyer, A. G.; Sydykova, D. K.; Spielman, S. J.; Wilke, C. O. Maximum Allowed Solvent Accessibilites of Residues in Proteins. PLoS ONE 2013, 8(11), e80635. doi:10.1371/journal.pone.0080635

Integration Libraries

• Gemmi: Wojdyr, M. GEMMI: A Library for Structural Biology. J. Open Source Softw. 2022, 7(73), 4200. doi:10.21105/joss.04200
• BioPython: Cock, P. J. A. et al. Biopython: Freely Available Python Tools for Computational Molecular Biology and Bioinformatics. Bioinformatics 2009, 25(11), 1422–1423. doi:10.1093/bioinformatics/btp163
• Biotite: Kunzmann, P.; Hamacher, K. Biotite: A Unifying Open Source Computational Biology Framework in Python. BMC Bioinformatics 2018, 19, 346. doi:10.1186/s12859-018-2367-z

License

MIT