analyze-sasa

name	analyze-sasa
description	Compute solvent-accessible surface area on a structure or MD trajectory with molecular-simulations — absolute per-residue SASA (Shrake-Rupley) and RelativeSASA fractional exposure (0=buried, 1=exposed). Use when measuring residue burial/exposure, finding solvent-exposed surfaces or buried interfaces, or quantifying conformational changes in accessibility over a trajectory.

SASA analysis

molecular_simulations.analysis.SASA and RelativeSASA implement the Shrake-Rupley algorithm as MDAnalysis AnalysisBase classes, so they iterate a trajectory and accept the standard run(start=, stop=, step=) slicing. Results are per-residue and averaged over the analyzed frames.

Absolute SASA

import MDAnalysis as mda
from molecular_simulations.analysis import SASA

u = mda.Universe('system.prmtop', 'prod.dcd')
sasa = SASA(u.select_atoms('protein'), probe_radius=1.4, n_points=256)
sasa.run()                       # or run(start=100, stop=500, step=10)
print(sasa.results.sasa)         # ndarray, one value per residue (Ų)

Arguments:

• ag — an MDAnalysis AtomGroup (not a whole Universe).
• probe_radius=1.4 — water probe radius in Å (standard).
• n_points=256 — points per atomic sphere; higher = more accurate but slower.

Requirements / gotchas:

• The Universe must expose an elements topology attribute — atomic radii come from MDAnalysis vdwradii keyed on element. AMBER .prmtop provides elements; a bare PDB may not (ValueError if missing).
• The AtomGroup must not be an UpdatingAtomGroup (e.g. avoid select_atoms(..., updating=True)) — raises TypeError.
• SASA is summed per residue using resid (1-indexed); make sure your selection is contiguous/sensible for the residue indexing you expect.

Relative SASA (fractional exposure)

RelativeSASA adds results.relative_area — each residue's SASA divided by its maximum accessible area, computed in a tripeptide context so the amide linkage and neighbors aren't over-counted. Values run 0 (buried) → 1 (exposed).

from molecular_simulations.analysis import RelativeSASA

u = mda.Universe('system.prmtop', 'prod.dcd')
rsasa = RelativeSASA(u.select_atoms('protein'))
rsasa.run()
print(rsasa.results.sasa)            # absolute SASA (Ų)
print(rsasa.results.relative_area)   # fractional exposure, 0..1

Extra requirement: the Universe needs a bonds attribute (the tripeptide reference uses a byres (bonded resindex ...) selection). AMBER topologies include bonds. If you loaded a structure without bonds, add them first, e.g. u.guess_bonds() or load alongside a topology that defines connectivity.

Single structure (no trajectory)

Load just the topology/coordinate pair and run() over the single frame:

u = mda.Universe('system.prmtop', 'system.inpcrd')
SASA(u.select_atoms('protein')).run()

Tips

• Subsample long trajectories with run(step=N) — SASA is the per-frame bottleneck (KDTree over n_points per atom).
• To map values back to residues, zip results.sasa with ag.residues.resids / .resnames.
• For interface burial, compute SASA of each partner alone vs. the complex and take the difference (buried surface area).
• Parallelize across many trajectories with the parsl-hpc skill — wrap the SASA(...).run() call in a CPU @python_app.

See also: analyze-ipsae (interface scoring of predicted complexes).