Analytical symmetry detection in protein assemblies. II. Dihedral and cubic symmetries

Guillaume Pagès, Sergei GrudininJournal of Structural Biology 2018 · DOI

TL;DR

This paper presents a novel computational method, AnAnaS, for accurately detecting dihedral and cubic symmetries in protein assemblies. This advancement is crucial for understanding the complex structural organization of proteins and has been validated on a large dataset from the Protein Data Bank (PDB), revealing potential inconsistencies in existing annotations.

What Problem Does This Paper Address?

Protein assemblies often exhibit high-order symmetries (like dihedral and cubic) that are fundamental to their function and stability. However, accurately detecting and analyzing these symmetries has been a significant computational challenge. Existing methods often struggle with the complexity of these symmetries, leading to potential inaccuracies in symmetry annotations. This paper aims to bridge this gap by providing a robust and precise method for identifying dihedral and cubic symmetries, thereby improving our understanding of protein structure and function.

Methods

The authors extend their previous work on cyclic symmetry detection to handle higher-order symmetries, specifically dihedral and cubic groups. Their approach involves:

  • Symmetry Measure: A root mean square deviation (RMSD) based measure is used to quantify the quality of a detected symmetry.
  • Heuristic Subunit Identification: A novel heuristic algorithm is introduced to identify "subunits" (the fundamental repeating units) within an assembly, even when these subunits are composed of multiple protein chains. This involves a penalty function based on sequence and structural alignments, followed by a clustering algorithm.
  • Analytical Optimization: The core of the method lies in an analytical approach to find the best symmetry axes. This involves expressing the problem using quaternion arithmetic and solving a constrained quadratic optimization problem, which is then reduced to a 2D trust-region subproblem solvable analytically. This ensures high precision and computational efficiency.
  • Group Representation: The paper utilizes a dual representation of symmetry groups, mapping permutations to rotations, which aids in the analytical solution.

Datasets

The primary dataset used for evaluation is the Protein Data Bank (PDB). The authors exhaustively analyzed approximately 51,358 symmetrical assemblies from the PDB, categorized into cyclic, dihedral, and cubic groups.

Key Insights and Ideas

  • Unified Approach for High-Order Symmetries: The paper successfully extends symmetry detection beyond simple cyclic symmetries to encompass the more complex dihedral and cubic groups, a significant step for structural biology.
  • Robust Subunit Identification: The heuristic method for identifying multi-chain subunits is a key innovation, allowing the analysis of more complex protein assemblies than previously possible.
  • Analytical Precision: The reliance on analytical solutions for optimization ensures a high degree of precision in symmetry axis determination, overcoming limitations of purely numerical methods.
  • PDB Annotation Verification: The exhaustive analysis of the PDB revealed that in a small but significant percentage of cases (1.6%), their method detected a higher symmetry group than what was annotated, highlighting potential inaccuracies in existing databases and the utility of their tool for quality control.
  • Size-Independence of Symmetry Imperfection: Interestingly, the study found no clear correlation between the size of an assembly and the degree of its symmetry imperfection, with larger assemblies often being more regularly organized.

Key Results

  • The AnAnaS method was implemented and tested on 51,358 PDB structures, demonstrating its robustness and efficiency.
  • In 1.6% of cases, AnAnaS identified a higher symmetry group than the PDB annotation.
  • The method achieved high accuracy, with RMSD values typically in the range of 0.2-0.5 Å for non-perfectly symmetrical assemblies.
  • The computational time scales favorably with assembly size compared to other methods.

Limitations and Future Work

  • Computational Bottlenecks: For some very complex structures (e.g., icosahedral symmetries), the discrete optimization step could lead to memory issues, causing the program to consider them asymmetric.
  • Handling of Missing/Extra Chains: The current program has limitations in handling structures with missing or additional chains that deviate from the expected number for a given symmetry group.
  • Thresholds for Symmetry: The definition of symmetry relies on RMSD thresholds (e.g., < 7 Å and < half the radius of gyration), which might exclude some borderline cases.
  • Future Work: The authors suggest that their method can be used to further investigate the structural organization of protein assemblies and to verify symmetry annotations in databases. The availability of the method as a standalone executable and a SAMSON module opens avenues for broader application and integration into existing bioinformatics workflows.

Summary updated

BibTeX
@article{pages2018analytical,
  author = {Guillaume Pagès and Sergei Grudinin},
  title = {{Analytical symmetry detection in protein assemblies. II. Dihedral and cubic symmetries}},
  journal = {Journal of Structural Biology},
  year = {2018},
  doi = {10.1016/j.jsb.2018.05.005},
}

Loading related content...

Figure