Summary:

Introduces the core engine module, a comprehensive framework for executing complex molecular modeling algorithms like sidechain placement and protein design. The engine is built around a configurable, stateful, and observable architecture. It includes a flexible configuration system with fluent builders, a state manager for tracking optimal solutions, task-based execution logic for modularity, and performance optimizations such as an Empty Lattice (EL) energy cache and optional parallelism via rayon. This work also refactors the forcefield module, decoupling topology and charge parameterization from the main runtime flow, as these are now handled during the pre-processing of input structures and rotamer libraries.

Changes:

• Implemented Core Engine Framework:

  • config.rs: Introduced configuration structs (PlacementConfig, DesignConfig, AnalyzeConfig) with fluent builders for type-safe and ergonomic setup of computational tasks.
  • state.rs: Developed OptimizationState and Solution structs. It uses a BinaryHeap to efficiently manage and track the top N solutions during an optimization run.
  • context.rs: Created a Context struct to centralize access to the system, configuration, and libraries. It includes a robust residue selection resolver that can handle lists, ligand binding sites (using a kiddo k-d tree for fast spatial lookups), and exclusions.
  • progress.rs: Implemented a callback-based ProgressReporter to provide feedback on long-running operations.
  • error.rs: Defined a comprehensive EngineError enum for structured error handling across the engine.

• Developed Algorithmic Tasks (tasks/):

  • el_energy.rs: Implemented the Empty Lattice (EL) energy calculation task to pre-compute and cache the interaction energy of each rotamer with the static protein backbone and environment.
  • doublet_optimization.rs: Implemented a pairwise rotamer optimization task, which is a key component of many sidechain packing algorithms.
  • clash_detection.rs: Provided a task to identify sterically clashing residue pairs based on a user-defined energy threshold.
  • total_energy.rs: Implemented a task to compute the total energy of a given system state by combining cached EL energies and pairwise interaction energies.

• Created Rotamer Placement Logic (placement.rs):

  • Implemented place_rotamer_on_system, which uses the Kabsch algorithm (via SVD) to perform a rigid-body alignment of a rotamer onto the backbone anchor atoms of a target residue.

• Refactored Forcefield Parameterization:

  • Simplified the Forcefield struct and Parameterizer by removing topology and charge assignment logic. This responsibility is now shifted to pre-processing steps (e.g., rotamer library loading), making the runtime parameterization focused on applying delta values and VDW properties.

• Enhanced Performance and Added Dependencies:

  • Added rayon under a parallel feature flag to enable parallel execution of computationally intensive tasks like EL energy calculation and clash detection.
  • Added kiddo for efficient k-d tree based nearest-neighbor searches, used in ligand binding site selection.
  • Added itertools for more complex iteration patterns.

• Minor Enhancements:

  • RotamerLibrary: Added include_system_conformations to allow including the initial sidechain conformation in the set of rotamers to be sampled.
  • EnergyTerm: Introduced a dedicated struct for energy components (vdw, coulomb, hbond) with Add and AddAssign implementations for cleaner energy arithmetic.