How broadinstitute/delphy works
Main System Components
a) Core Engine Layer:
• Subdivide into these key modules (based on the core/ directory):
• Phylogenetic Tree Management (phylo_tree.h, tree.h)
• Evolution Models (evo_model.h, evo_hky.h)
• Coalescent Processing (coalescent.h, scalable_coalescent.h)
• Input/Output Handling (io.h, api.h)
b) Interface Layer:
• Multiple entry points:
• Command-line interface (tools/delphy.cpp)
• UI interface (tools/delphy_ui.cpp)
• WebAssembly interface (tools/delphy_wasm.cpp)
c) External Dependencies:
• Third-party components (from third-party/):
• abseil-cpp
• cppcoro
• flatbuffers
• cxxopts
Data Flow and Interactions
• Input flow: Show how data enters through different interfaces
• Processing flow: Illustrate how data moves through the core engine
• Output flow: Show how results are returned to users
Specific Components to Highlight
a) EMAT (Explicit Mutation-Annotated Trees) Processing:
• Mutation annotation handling (mutations.h)
• Tree probing mechanisms (tree_prober.h)
• Site state management (site_states_tree_prober.h)
b) Algorithmic Components:
• Relationship between:
• SPR (Subtree Pruning and Regrafting) operations (spr_move.h, spr_study.h)
• Coalescent calculations (scalable_coalescent.h, very_scalable_coalescent.h)
• Phylogenetic tree calculations (phylo_tree_calc.h)
Technical Details
Annotations for:
• File formats supported
• Key algorithms implemented
• Performance optimization points
• Scaling capabilities
• The modular nature of the core engine
• The multiple interface options
• The separation between computational and I/O components
• The relationship with external tools like UShER
Keep the abstraction level focused on the architectural components while providing enough detail to understand the unique aspects of Delphy's EMAT-based approach and its scaling capabilities for genomic epidemiology datasets.