AI-Guided Lipid Nanoparticle Design for mRNA & Gene Delivery: Formulation Prediction and Optimization

Aim

Workshop Objectives

Workshop Structure

📅 Day 1: Lipid Nanoparticle and mRNA Delivery Fundamentals

• Introduction to lipid nanoparticles for mRNA and gene delivery
• Key components of LNP formulations
• Ionisable lipids and endosomal escape
• Helper phospholipids and membrane stability
• Roles of cholesterol and PEG lipids
• Encapsulation efficiency, particle size, and surface properties
• Biological barriers affecting mRNA delivery
• Cellular uptake, intracellular trafficking, and cargo release
• Introduction to lipid and chemical databases
• Overview of LNP formulation datasets

🛠️ Hands-on:

• Explore lipid structures and properties using LIPID MAPS and PubChem
• Retrieve molecular information for selected LNP components
• Examine formulation variables such as lipid ratios, particle size, and encapsulation efficiency
• Prepare a structured lipid-formulation dataset for analysis

🧰 Tools Covered:

LIPID MAPS, PubChem, formulation datasets, and data-curation templates

📅 Day 2: AI-Based Formulation Prediction

• Collection and organisation of lipid-formulation data
• Molecular representation and descriptor generation
• Data cleaning and missing-value treatment
• Encoding formulation and experimental variables
• Feature selection and dimensionality reduction concepts
• Delivery-efficiency and encapsulation prediction
• Regression and classification approaches for LNP modelling
• Comparison of machine-learning algorithms
• Model validation and performance metrics
• Interpretation of formulation–performance relationships

🛠️ Hands-on:

• Generate lipid molecular descriptors using RDKit
• Clean and preprocess formulation data using Pandas
• Train delivery-performance models using scikit-learn and XGBoost
• Compare model performance using suitable evaluation metrics
• Identify influential molecular and formulation features

🧰 Tools Covered:

Python, Google Colab, RDKit, Pandas, scikit-learn, XGBoost, and formulation datasets

📅 Day 3: LNP Optimisation, Interpretation and Candidate Selection

• Multi-parameter optimisation of LNP formulations
• Balancing delivery efficiency, encapsulation, stability, and toxicity
• Hyperparameter and formulation-space optimisation
• Prediction of delivery performance and potential toxicity
• SHAP-based model interpretation
• Identification of important lipid and formulation variables
• Candidate formulation scoring and ranking
• Data visualisation for formulation comparison
• Dashboard design for LNP prediction workflows
• Research and pharmaceutical-development applications

🛠️ Hands-on:

• Optimise model and formulation parameters using Optuna
• Interpret delivery predictions using SHAP
• Rank candidate LNP formulations using multiple performance criteria
• Create interactive formulation visualisations using Plotly
• Develop a basic LNP prediction dashboard prototype using Streamlit

🧰 Tools Covered:

Python, XGBoost, Optuna, SHAP, Plotly, Streamlit, and candidate-ranking templates

Who Should Enrol?

Important Dates

Workshop Outcomes