DeepTraCDR/Scenario1/Random/train_random.py

import argparse
import numpy as np
import pandas as pd
import torch
from sklearn.model_selection import KFold
from DeepTraCDR_model import DeepTraCDR, Optimizer
from utils import evaluate_auc
from data_sampler import RandomSampler
from data_loader import load_data


def parse_arguments():
    """
    Parses command-line arguments for the DeepTraCDR model.

    Returns:
        Parsed arguments containing model and training configurations.
    """
    parser = argparse.ArgumentParser(description="DeepTraCDR: Graph-based Cell-Drug Interaction Prediction")
    parser.add_argument('-device', type=str, default="cuda:0" if torch.cuda.is_available() else "cpu",
                        help="Device to run the model on (cuda:0 or cpu)")
    parser.add_argument('-data', type=str, default='ccle', help="Dataset to use (default: ccle)")
    parser.add_argument('--wd', type=float, default=1e-4, help="Weight decay for optimizer")
    parser.add_argument('--layer_size', nargs='+', type=int, default=[512], help="Layer sizes for the model")
    parser.add_argument('--gamma', type=float, default=15, help="Gamma parameter for decoder")
    parser.add_argument('--epochs', type=int, default=1000, help="Number of training epochs")
    parser.add_argument('--test_freq', type=int, default=50, help="Frequency of evaluation during training")
    parser.add_argument('--lr', type=float, default=0.0005, help="Learning rate for optimizer")
    return parser.parse_args()


def main():
    """Main function to execute the DeepTraCDR training and evaluation pipeline."""
    args = parse_arguments()

    # Load dataset
    full_adj, drug_fingerprints, exprs, null_mask, pos_num, args = load_data(args)
    
    # Log data shapes for debugging
    print(f"Original adj_mat shape: {full_adj.shape}")
    print("\n--- Data Shapes ---")
    print(f"Expression data shape: {exprs.shape}")
    print(f"Null mask shape: {null_mask.shape}")

    # Convert adjacency matrix to torch tensor if necessary
    if isinstance(full_adj, np.ndarray):
        full_adj = torch.from_numpy(full_adj).float()
    print(f"Converted adj_mat shape: {full_adj.shape}")

    # Initialize k-fold cross-validation
    k = 5
    n_kfolds = 5
    all_metrics = {
        'auc': [], 'auprc': [], 'precision': [], 'recall': [], 'f1_score': []
    }

    # Perform k-fold cross-validation
    for n_kfold in range(n_kfolds):
        kfold = KFold(n_splits=k, shuffle=True, random_state=n_kfold)
        for fold, (train_idx, test_idx) in enumerate(kfold.split(np.arange(pos_num))):
            # Initialize data sampler
            sampler = RandomSampler(full_adj, train_idx, test_idx, null_mask)

            # Initialize model
            model = DeepTraCDR(
                adj_mat=full_adj,
                cell_exprs=exprs,
                drug_finger=drug_fingerprints,
                layer_size=args.layer_size,
                gamma=args.gamma,
                device=args.device
            )

            # Initialize optimizer
            opt = Optimizer(
                model=model,
                train_data=sampler.train_data,
                test_data=sampler.test_data,
                test_mask=sampler.test_mask,
                train_mask=sampler.train_mask,
                adj_matrix=full_adj,
                evaluate_fun=evaluate_auc,
                lr=args.lr,
                wd=args.wd,
                epochs=args.epochs,
                test_freq=args.test_freq,
                device=args.device
            )

            # Train model and collect metrics
            true, pred, best_auc, best_auprc, best_precision, best_recall, best_f1 = opt.train()
            
            # Store metrics
            all_metrics['auc'].append(best_auc)
            all_metrics['auprc'].append(best_auprc)
            all_metrics['precision'].append(best_precision)
            all_metrics['recall'].append(best_recall)
            all_metrics['f1_score'].append(best_f1)

            print(f"Fold {n_kfold * k + fold + 1}: AUC={best_auc:.4f}, AUPRC={best_auprc:.4f}, "
                  f"Precision={best_precision:.4f}, Recall={best_recall:.4f}, F1-Score={best_f1:.4f}")

    # Compute and display final metrics
    print("\nFinal Average Metrics:")
    for metric, values in all_metrics.items():
        mean = np.mean(values)
        std = np.std(values)
        print(f"{metric.upper()}: {mean:.4f} ± {std:.4f}")


if __name__ == "__main__":
    torch.set_float32_matmul_precision('high')
    main()