configs/paths.yaml

adsorption_models:
  root: "data/adsorption"
  benzene_model: "data/adsorption/benzene/rf_only_seed8_best_model_benzene.pkl"
  benzene_base_model: "data/adsorption/benzene/best_model_benzene.pkl"
  benzene_scaler: "data/adsorption/benzene/scaler_benzene.pkl"
  toluene_model: "data/adsorption/toluene/best_model_toluene.pkl"
  toluene_scaler: "data/adsorption/toluene/scaler_toluene.pkl"
  benzene_target_dim: 520
  toluene_target_dim: 584
  benzene_training_data: "data/adsorption/benzene/sine_matrix_benzene_IQR_2_modeling_cleaned.csv"
  toluene_training_data: "data/adsorption/toluene/sine_matrix_toluene_modeling_cleaned_toluene.csv"

toxicity_models:
  root: "data/toxicity"
  model_dir: "trained_toxicity_models"
  endpoints:
    - name: "LC50"
      organism: "Pimephales promelas"
      duration: "96h"
      training_sdf: "data/toxicity/training/LC50_training.sdf"
      training_csv: "data/toxicity/training/LC50_training.csv"
    - name: "LC50DM"
      organism: "Daphnia magna"
      duration: "48h"
      training_sdf: "data/toxicity/training/LC50DM_training.sdf"
      training_csv: "data/toxicity/training/LC50DM_training.csv"
    - name: "IGC50"
      organism: "Tetrahymena pyriformis"
      duration: "48h"
      training_sdf: "data/toxicity/training/IGC50_training.sdf"
      training_csv: "data/toxicity/training/IGC50_training.csv"
    - name: "IBC50"
      organism: "Vibrio fischeri"
      duration: "15min"
      training_sdf: "data/toxicity/training/IBC50_training.sdf"
      training_csv: "data/toxicity/training/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