CYP-MAP

Overview

CYP-MAP is an advanced computational model designed to accurately predict Sites of Metabolism (SoMs) for small molecules, particularly those metabolized by cytochrome (CYP) enzymes. By integrating a multi-level Graph Neural Network (GNN) model with the largest curated CYP-mediated metabolism database to date, CYP-MAP provides unprecedented accuracy in metabolic site prediction. This powerful tool enables researchers in drug discovery, medicinal chemistry, and toxicology to better understand and optimize drug metabolism, ultimately improving drug efficacy and safety.

---

Key Features and Innovations

1. Comprehensive CYP-Mediated Metabolism Database

CYP-MAP is built upon the most extensive database of CYP-mediated SoMs, integrating data from:

• EBoMD, DrugBank, and other publicly available resources.
• Extensive literature searches, ensuring broad chemical diversity and high-quality experimental validation.

This dataset enables CYP-MAP to predict metabolic reactions across multiple CYP isoforms, ensuring a robust and accurate representation of metabolic pathways.

2.GNN Model

CYP-MAP employs a novel multi-level GNN architecture that enhances metabolic site prediction by considering three levels of molecular representation:

• Atom-Level Representation: Identifies precise atomic sites of metabolism.
• Bond-Level Representation: Captures bond transformations during CYP-mediated reactions.
• Whole-Molecule Representation: Evaluates global molecular properties to determine metabolic susceptibility.

By leveraging these interconnected representations, CYP-MAP achieves state-of-the-art performance, significantly surpassing traditional rule-based models and previous machine learning approaches.

3. Substrate, SoM, and Reaction Type Prediction

Unlike conventional SoM prediction models, which only identify metabolic sites, CYP-MAP provides a more comprehensive analysis, predicting:

• CYP Substrate Specificity: Determines whether a molecule will be metabolized by a specific CYP enzyme.
• Sites of Metabolism (SoMs): Predicts the most likely metabolic transformation sites.
• Reaction Types: Classifies the type of metabolic transformation (e.g., hydroxylation, dealkylation, oxidation).

This integrated approach allows researchers to align metabolic site predictions with expected reaction outcomes, improving metabolite structure prediction.

Address

06974 서울특별시 동작구 흑석로 84
중앙대학교 102관(약학대학 및 R&D센터) 612호

84 Heukseok-ro, Dongjak-gu, Seoul 06974
Building 102 (College of Pharmacy/R&D Center) #612

Phone: +82-2-820-5674

If you encounter any issues or have any questions, please feel free to contact us at the email address below.

Email: yoonjilee@cau.ac.kr