configs/paths.yaml.example

# Copy this file to paths.yaml and fill in your local model paths
adsorption_models:
  root: "<YOUR_MODEL_PATH>/sine_matrix"
  benzene_model: "<YOUR_MODEL_PATH>/sine_matrix/Benzene/tuning_seed8/rf_only_seed8_best_model_benzene.pkl"
  benzene_base_model: "<YOUR_MODEL_PATH>/sine_matrix/Benzene/best_model_benzene.pkl"
  benzene_scaler: "<YOUR_MODEL_PATH>/sine_matrix/Benzene/scaler_benzene.pkl"
  toluene_model: "<YOUR_MODEL_PATH>/sine_matrix/Toluene/Toluene/best_model_toluene.pkl"
  toluene_scaler: "<YOUR_MODEL_PATH>/sine_matrix/Toluene/Toluene/scaler_toluene.pkl"
  benzene_target_dim: 520
  toluene_target_dim: 584
  benzene_training_data: "<YOUR_MODEL_PATH>/sine_matrix/Benzene/sine_matrix_benzene_IQR_2_modeling_cleaned.csv"
  toluene_training_data: "<YOUR_MODEL_PATH>/sine_matrix/Toluene/Toluene/sine_matrix_toluene_modeling_cleaned_toluene.csv"

toxicity_models:
  root: "<YOUR_MODEL_PATH>/AquaTox"
  model_dir: "<YOUR_PROJECT_PATH>/safe_mats_demo/trained_toxicity_models"
  endpoints:
    - name: "LC50"
      organism: "Pimephales promelas"
      duration: "96h"
      training_sdf: "<YOUR_MODEL_PATH>/AquaTox/data/LC50/LC50_training.sdf"
      training_csv: "<YOUR_MODEL_PATH>/AquaTox/data/LC50/LC50_training.csv"
    - name: "LC50DM"
      organism: "Daphnia magna"
      duration: "48h"
      training_sdf: "<YOUR_MODEL_PATH>/AquaTox/data/LC50DM/LC50DM_training.sdf"
      training_csv: "<YOUR_MODEL_PATH>/AquaTox/data/LC50DM/LC50DM_training.csv"
    - name: "IGC50"
      organism: "Tetrahymena pyriformis"
      duration: "48h"
      training_sdf: "<YOUR_MODEL_PATH>/AquaTox/data/IGC50/IGC50_training.sdf"
      training_csv: "<YOUR_MODEL_PATH>/AquaTox/data/IGC50/IGC50_training.csv"
    - name: "IBC50"
      organism: "Vibrio fischeri"
      duration: "15min"
      training_sdf: "<YOUR_MODEL_PATH>/AquaTox/data/IBC50/IBC50_training.sdf"
      training_csv: "<YOUR_MODEL_PATH>/AquaTox/data/IBC50/IBC50_training.csv"
  fingerprint:
    type: "morgan"
    radius: 2
    n_bits: 2048
  model_params:
    n_estimators: 500
    max_depth: null
    min_samples_leaf: 2
    random_state: 42
    test_size: 0.2