Protein Structure Search Prototype

Protein Structure Search Prototype compares electron density volumes using 3D Zernike moments to detect and retrieve similar protein structures at the oligomeric assembly level for structural comparison and classification.

Key Features:

• Electron density volume comparison: Direct comparison of electron density volumes instead of relying on atomic coordinates or chain connectivity.
• Avoidance of atomic-coordinate dependence: Circumvents limitations associated with atomic coordinates, polypeptide chain connectivity, and atom-level topological permutations.
• 3D Zernike decomposition: Uses 3D Zernike polynomials to decompose volumetric functions into compact vectors known as Zernike moments.
• Rotation-invariant descriptors: Generates rotation-invariant descriptions of protein shapes via Zernike moments for orientation-independent comparison.
• Normalization and orientation: Applies a normalization procedure to orient and normalize electron density volumes prior to comparison.
• Handling of oligomeric assemblies: Designed to detect similarity and alignments at the scale of oligomeric assemblies, including cases with subunit rearrangements.
• Retrieval and alignment: Enables rapid retrieval of matching entries from the Protein Data Bank (PDB) or user-provided inputs and provides alignments.

Scientific Applications:

• Structure similarity detection: Identification of similar protein structures across atomic and assembly-level variations.
• Classification of assemblies: Classification of diverse protein assemblies based on volumetric shape descriptors.
• Evolutionary analysis: Detection of structural patterns relevant to evolutionary relationships among proteins.
• Functional inference: Support for inferring functional relationships from shape-based structural similarity.
• Large-assembly searches: Structural searches at the scale of the PDB focused on large and oligomeric protein assemblies.

Methodology:

Direct comparison of electron density volumes using 3D Zernike polynomials to compute Zernike moments, application of a normalization procedure to orient and produce rotation-invariant descriptors, and retrieval/alignment against entries from the Protein Data Bank (PDB) or user inputs.