Skip to content

PLAN.md

Latest commit

 

History

History
77 lines (58 loc) · 3.12 KB

PLAN.md

File metadata and controls

77 lines (58 loc) · 3.12 KB
Raw

Implementation Plan

Module dependency graph

Day 1                         Day 2                         Day 3
──────                        ──────                        ──────
A.aa_md ──┐                   A.continuum ──┐               All figures
           ├─ A.cg_bridge ──┘                ├─ A validation               
           │                                │                              
B (独立) ──────────────────────────────────┘               B figure
C (独立) ──────────────────────────────────┘               C figure
D ─────── (needs A completed parameter set) ─┘             D figure

Module interfaces (data contracts)

A: Three-scale pipeline

A.aa_md → A.cg_bridge: File: data/aa_features.csv Columns: residue_i, residue_j, contact_prob, avg_distance, hbond_lifetime Generating code: modules/A_triScale/aa_md/extract_features.py

A.cg_bridge → A.continuum: File: data/cg_kinetics.csv Columns: condition (conc/salt), D_diffusion, k_on_dimer, k_off_dimer Generating code: modules/A_triScale/cg_bridge/extract_kinetics.py

A.continuum → validation: File: data/cluster_distribution.csv Columns: time, cluster_size, concentration

B: Adaptive path exploration

Input: none (self-contained Müller-Brown potential) Output: outputs/figures/fig4_paths.png

C: RL optimization

Input: none (self-contained HP lattice) Output: outputs/figures/fig5_rl_bo_comparison.png

D: BO prediction

Input: data/aa_features.csv + data/cg_kinetics.csv (from Module A) Output: outputs/figures/fig6_bo.png

Implementation order

Day 1 (hours 0-10)

  1. modules/A_triScale/aa_md/run_md.py — OpenMM peptide simulation
  2. modules/A_triScale/aa_md/extract_features.py — AA trajectory analysis
  3. modules/A_triScale/cg_bridge/train_ml.py — MLP for AA→CG mapping
  4. modules/A_triScale/cg_bridge/run_cg.py — CG simulation with learned params

Day 2 (hours 10-20)

  1. modules/A_triScale/continuum/solve_ode.py — Smoluchowski solver
  2. modules/A_triScale/validate.py — Cross-scale validation
  3. modules/B_pathExploration/run_sampling.py — Müller-Brown path sampling
  4. modules/C_rlOptimization/train_rl.py — HP lattice RL
  5. modules/D_boPrediction/run_bo.py — XGBoost + BO

Day 3 (hours 20-28)

  1. All figure scripts in outputs/figures/
  2. Report writing

Figure output mapping

Figure Script Module PPT mapping
fig1_triScale_overview.png plot_triScale.py A Multi-scale coupling diagram
fig2_fes_slowVars.png plot_fes.py A ML enhanced sampling + FES
fig3_parameter_validation.png plot_validation.py A Parameter transfer validation
fig4_adaptive_paths.png plot_paths.py B Adaptive path exploration
fig5_rl_convergence.png plot_rl.py C RL vs BO convergence
fig6_bo_optimization.png plot_bo.py D BO optimization + feature importance