Antibody–Drug Conjugates: Target Selection, Linker–Payload Design & AI Toxicity Prediction

Aim

Workshop Objectives

Workshop Structure

📅 Day 1: ADC Biology, Target Selection & Precision-Therapeutic Profiling

• Fundamentals of Antibody–Drug Conjugates and targeted therapeutics
• ADC components: antibody, linker, payload, and conjugation strategy
• Target binding, internalization, linker cleavage, and payload release
• Ideal ADC target characteristics and tumor-specific antigen expression
• Genomic alterations, target prevalence, and normal-tissue off-target risk
• Precision-oncology applications and target-prioritization principles

🛠️ Hands-on:

• Explore candidate targets using cBioPortal
• Assess tumor and normal-tissue expression using Human Protein Atlas
• Compare genomic prevalence, selectivity, and biological relevance
• Prepare a ranked shortlist of potential ADC targets

🧰 Tools Covered:

cBioPortal, Human Protein Atlas (HPA), cancer genomics datasets, and target-ranking templates

📅 Day 2: Linker–Payload Design, Chemical Space & Molecular Profiling

• Cleavable and non-cleavable linker strategies
• Payload-release mechanisms and major cytotoxic payload classes
• Microtubule inhibitors, DNA-damaging agents, and topoisomerase inhibitors
• Potency, permeability, bystander effect, and therapeutic window
• Drug-to-antibody ratio and conjugation considerations
• Structure–activity and structure–property relationships
• Molecular descriptors influencing payload toxicity and developability

🛠️ Hands-on:

• Retrieve cytotoxic compounds and activity data from ChEMBL
• Generate molecular descriptors using RDKit
• Analyse molecular weight, logP, TPSA, and hydrogen-bond properties
• Compare payload chemical profiles and visualize chemical space
• Prepare an AI-ready payload dataset

🧰 Tools Covered:

ChEMBL, RDKit, Python, pandas, Google Colab, and molecular-property datasets

📅 Day 3: AI Toxicity Prediction, Explainable AI & ADC Candidate Prioritization

• On-target, off-target, and payload-related toxicity
• Linker instability and premature payload release
• Machine learning for molecular toxicity prediction
• Feature engineering and XGBoost-based modelling
• Model validation and performance evaluation
• Explainable AI using SHAP
• Multi-parameter ADC prioritization
• Balancing target selectivity, payload potency, safety, and developability
• Translational considerations in ADC development

🛠️ Hands-on:

• Train an XGBoost model using molecular descriptors
• Evaluate toxicity-prediction performance
• Identify toxicity-driving features using SHAP
• Compare payload candidates by predicted safety
• Integrate target, payload, and toxicity data
• Rank preliminary ADC candidates for further development

🧰 Tools Covered:

Google Colab, Python, RDKit, XGBoost, SHAP, pandas, scikit-learn, and ADC-ranking templates

Who Should Enrol?

Important Dates

Workshop Outcomes