Representations of lipid nanoparticles using large language models for transfection efficiency prediction

Area of Science:

• Biotechnology
• Computational Chemistry
• Vaccine Development

Background:

• Lipid nanoparticles (LNPs) are crucial for mRNA vaccine delivery.
• LNP lipid structure significantly influences mRNA payload effectiveness.
• Key properties for optimization include biodegradability, synthetic accessibility, and transfection efficiency.

Purpose of the Study:

• To develop and validate computational models for virtual screening of LNPs.
• To identify LNPs with enhanced transfection efficiency for mRNA delivery.
• To accelerate the optimization of LNP formulations for improved vaccine efficacy.

Main Methods:

• Utilized Simplified Molecular-Input Line-Entry System (SMILES) strings of lipids as input for a large language model (LLM).
• Employed LLM-generated embeddings as features for a gradient-boosting classifier.
• Developed a predictive model for assessing LNP properties and transfection efficiency.

Main Results:

• The developed model demonstrates high accuracy in predicting lipid properties relevant to LNP performance.
• The LLM-based approach enables efficient virtual screening of potential LNP candidates.
• Predicted improvements in transfection efficiency and potential for reduced experimental costs.

Conclusions:

• The LLM-driven computational approach offers a powerful tool for optimizing LNP design for mRNA vaccines.
• This method can significantly reduce the time and resources required for LNP development.
• The findings pave the way for more effective and cost-efficient mRNA vaccine delivery systems.