Fast Computation of Exact Symmetry-Corrected RMSD of Conformers

Area of Science:

• Computational Chemistry
• Structural Biology
• Bioinformatics

Background:

• Accurate comparison of molecular conformers is crucial for understanding biological processes.
• Existing methods for atomic correspondence and RMSD calculation face challenges with computational efficiency and topological correctness, especially for large molecules like proteins.

Purpose of the Study:

• To introduce a novel algorithm, Hierarchical Neighborhood of Atoms (HNA) partitioning, for ultrafast and exact symmetry-corrected atomic correspondence and RMSD computation.
• To address the limitations of existing topology-unaware and graph isomorphism methods in handling molecular conformer comparisons.

Main Methods:

• Developed a recursive classification of chemically equivalent atoms (HNA partitioning).
• Implemented a hierarchical decomposition of the assignment problem to reduce computational complexity.
• Tested the algorithm on large datasets of protein conformers, comparing its performance against existing methods.

Main Results:

• Achieved reductions of up to 16 orders of magnitude in evaluated combinations compared to product-based approaches.
• Demonstrated 100% topologically correct assignments without timeouts across 1.4 million protein conformer pairs.
• Reported millisecond-scale execution times and 11-42× speedups over polynomial-time methods.

Conclusions:

• The HNA partitioning method provides a significant advancement in the speed and accuracy of molecular conformer comparison.
• This framework offers a robust solution for symmetry-consistent atomic property comparisons, canonical labeling, and force field parametrization.