A novel in silico platform combining data-driven and physics-based models for protein formulation developability...

Area of Science:

• Biopharmaceutical Development
• Computational Chemistry
• Protein Engineering

Background:

• Therapeutic protein development involves significant investment and faces high failure rates.
• Failures often stem from molecular instability and formulation challenges, particularly in early development stages.
• Limited early-stage drug substance necessitates alternative assessment methods like computational modeling.

Purpose of the Study:

• To present a novel in silico platform for assessing therapeutic protein formulation developability.
• To reduce financial and time risks associated with costly drug development failures.
• To provide a method that requires only the protein's primary sequence, bypassing the need for physical material.

Main Methods:

• Development of a novel in silico platform integrating data-driven and physics-based models.
• Utilized computational techniques including structure prediction, bioinformatics, machine learning, and molecular dynamics.
• Platform input is the primary amino acid sequence of the therapeutic protein.

Main Results:

• The platform identifies protein liabilities and outlying properties.
• It evaluates suitability for high-concentration formulations and screens formulation condition effects on self-interaction.
• The system suggests optimal formulation corridors for therapeutic proteins.

Conclusions:

• The developed in silico platform offers a powerful tool for early-stage assessment of protein drug candidates.
• It enables risk reduction by predicting formulation developability without physical material.
• Case studies demonstrate the platform's applicability and experimental validation in real-world scenarios.