CompDReAM

Drug repurposing framework for Parkinson disease using ChEMBL + DisGeNET. It covers a gene scoring pipeline, followed by machine learning regression models for pChEMBL values.

Machine Learning data

The contents of the machinelearning folder are too large for GitHub and are therefore hosted externally.

🔗 Access all machine learning files via Google Drive

---

Folder contents on Drive

• /v1, /v2, ..., /v5: Model versions (Random Forest, SVR, etc.)
• /vX/protbert/: Protein embeddings (e.g., X_prot.npy, X_combined.npy)
• training_dataset.csv, training_dataset_scoregda.csv: Datasets used in training
• rf_model.pkl, rf_metadata_summary.csv, feature_importance.png: Model metadata and visualisations

---

How to use these files (in Colab or locally)

from google.colab import drive
drive.mount('/content/drive')

# Example: Load combined features for v5
import numpy as np
X = np.load('/content/drive/MyDrive/CompDReAM/v5/X_combined.npy')

# Load trained model
import joblib
model = joblib.load('/content/drive/MyDrive/CompDReAM/v5/Random_Forest.pkl')

---