NMFF

NMFF performs flexible fitting of high-resolution macromolecular structures into low-resolution electron microscopy density maps by leveraging low-frequency normal modes derived from multi-resolution elastic network models to model and optimize conformational variability.

Key Features:

• Normal Mode Analysis: Computes low-frequency normal mode vectors to explore collective conformational variability of macromolecules.
• Multi-resolution Elastic Network Model: Uses a multi-resolution elastic network representation to generate mode vectors from different structural resolutions.
• Iterative Mode Combination and Deformation: Applies linear combinations of low-frequency modes iteratively to deform high-resolution structures toward target densities.
• Correlation-based Density Optimization: Maximizes the correlation between computed electron density from the flexible model and the experimental density map to guide fitting.
• Gradient-following Optimization in Collective Modes: Incorporates gradient-following techniques in collective normal modes to locally optimize the overall correlation coefficient.
• Multi-scale Structural Models: Supports normal mode searching on lower-resolution multi-scale models without requiring fully atomic input structures.

Scientific Applications:

• Flexible Docking: Quantitative flexible docking of macromolecular complexes into coarser electron density maps from techniques such as cryo-electron microscopy (cryo-EM).
• Conformational Change Modeling: Modeling and refinement of proteins and other biomolecules that undergo significant conformational changes.
• Structure Refinement of Macromolecular Assemblies: Rapid refinement of high-resolution structures into low-resolution density frameworks to improve agreement with experimental maps.

Methodology:

Constructs multi-resolution elastic network representations to compute low-frequency normal mode vectors; applies iterative linear combinations of these modes to deform models and maximize correlation between computed model density and experimental density maps; employs gradient-following optimization in collective normal modes; supports normal mode searching on lower-resolution multi-scale structural models.