Diffdock

complex_name,protein_path,ligand_description,protein_sequence
example_1,protein1.pdb,CC(=O)Oc1ccccc1C(=O)O,
example_2,,COc1ccc(C#N)cc1,MSKGEELFTGVVPILVELDGDVNGHKFSVSGEGEGDATYGKLTLKFICTTGKLPVPWPTLVTTFSYGVQCFSRYPDHMKQHDFFKSAMPEGYVQERTIFFKDDGNYKTRAEVKFEGDTLVNRIELKGIDFKEDGNILGHKLEYNYNSHNVYIMADKQKNGIKVNFKIRHNIEDGSVQLADHYQQNTPIGDGPVLLPDNHYLSTQSALSKDPNEKRDHMVLLEFVTAAGITHGMDELYK
example_3,protein3.pdb,ligand3.sdf,


# DiffDock Custom Inference Configuration Template
# Copy and modify this file to customize inference parameters

# Model paths (usually don't need to change these)
model_dir: ./workdir/v1.1/score_model
confidence_model_dir: ./workdir/v1.1/confidence_model
ckpt: best_ema_inference_epoch_model.pt
confidence_ckpt: best_model_epoch75.pt

# Model version flags
old_score_model: false  # Set to true to use original DiffDock instead of DiffDock-L
old_filtering_model: true

# Inference steps
inference_steps: 20  # Increase for potentially better accuracy (e.g., 25-30)
actual_steps: 19
no_final_step_noise: true

# Sampling parameters
samples_per_complex: 10  # Increase for difficult cases (e.g., 20-40)
sigma_schedule: expbeta
initial_noise_std_proportion: 1.46

# Temperature controls - Adjust these to balance exploration vs accuracy
# Higher values = more diverse predictions, lower values = more focused predictions

# Sampling temperatures
temp_sampling_tr: 1.17   # Translation sampling temperature
temp_sampling_rot: 2.06  # Rotation sampling temperature
temp_sampling_tor: 7.04  # Torsion sampling temperature (increase for flexible ligands)

# Psi angle temperatures
temp_psi_tr: 0.73
temp_psi_rot: 0.90
temp_psi_tor: 0.59

# Sigma data temperatures
temp_sigma_data_tr: 0.93
temp_sigma_data_rot: 0.75
temp_sigma_data_tor: 0.69

# Feature flags
no_model: false
no_random: false
ode: false  # Set to true to use ODE solver instead of SDE
different_schedules: false
limit_failures: 5

# Output settings
# save_visualisation: true  # Uncomment to save SDF files

# ============================================================================
# Configuration Presets for Common Use Cases
# ============================================================================

# PRESET 1: High Accuracy (slower, more thorough)
# samples_per_complex: 30
# inference_steps: 25
# temp_sampling_tr: 1.0
# temp_sampling_rot: 1.8
# temp_sampling_tor: 6.5

# PRESET 2: Fast Screening (faster, less thorough)
# samples_per_complex: 5
# inference_steps: 15
# temp_sampling_tr: 1.3
# temp_sampling_rot: 2.2
# temp_sampling_tor: 7.5

# PRESET 3: Flexible Ligands (more conformational diversity)
# samples_per_complex: 20
# inference_steps: 20
# temp_sampling_tr: 1.2
# temp_sampling_rot: 2.1
# temp_sampling_tor: 8.5  # Increased torsion temperature

# PRESET 4: Rigid Ligands (more focused predictions)
# samples_per_complex: 10
# inference_steps: 20
# temp_sampling_tr: 1.1
# temp_sampling_rot: 2.0
# temp_sampling_tor: 6.0  # Decreased torsion temperature

# ============================================================================
# Usage Example
# ============================================================================
# python -m inference \
#   --config custom_inference_config.yaml \
#   --protein_ligand_csv input.csv \
#   --out_dir results/


# DiffDock Confidence Scores and Limitations

This document provides detailed guidance on interpreting DiffDock confidence scores and understanding the tool's limitations.

## Confidence Score Interpretation

DiffDock generates a confidence score for each predicted binding pose. This score indicates the model's certainty about the prediction.

### Score Ranges

| Score Range | Confidence Level | Interpretation |
|------------|------------------|----------------|
| **> 0** | High confidence | Strong prediction, likely accurate binding pose |
| **-1.5 to 0** | Moderate confidence | Reasonable prediction, may need validation |
| **< -1.5** | Low confidence | Uncertain prediction, requires careful validation |

### Important Notes on