AI based Bispecific Antibody (BsAb) Optimization Service

Creative Biolabs specializes in optimizing bispecific antibodies (BsAb) using advanced AI technologies. By skillfully leveraging AI-based methodologies, we help to accelerate BsAb design, optimize their stability, and enhance their therapeutic potential. Our services focus on improving the efficacy, safety, and manufacturability of BsAbs, providing faster, more cost-effective solutions to global clients.

Introduction

AI refers to systems that mimic human intelligence, with machine learning (ML) and deep learning (DL) focusing on predictive modeling and data analysis. Therapeutic antibodies, including BsAbs, have proven effective in treating diseases, but their design and discovery remain complex and time-consuming. AI has significantly impacted antibody design and discovery. AI-based antibody engineering holds great promise for accelerating and optimizing BsAb development.

BsAb Structure Modeling Based on ML

ML techniques allow the prediction and optimization of the structure of BsAbs by analyzing large datasets of antibody sequences and structures. ML models can efficiently identify patterns in these datasets, helping to design BsAb constructs with optimal binding affinities and stability. These models also aid in predicting the behavior of the antibody under various conditions, thus enhancing its overall performance. Antibody structures have conventionally been predicted using either physics-based modeling or homology-based modeling, or a combined method.

BsAb Structure Modeling Based on DL

DL offers a more advanced approach for modeling BsAb structures by utilizing neural networks to predict the 3D configuration of antibodies. Through training on vast amounts of structural data, DL algorithms can provide highly accurate models of BsAbs. These models can capture complex interactions and predict the structural outcomes of modifications, enabling better design and optimization of BsAbs for specific therapeutic targets.

AI-Based Models for BsAb Structure Prediction and Design

AI-driven models for BsAb structure prediction and design utilize ML and DL techniques to simulate antibody-antigen interactions, predict structural stability, and suggest optimal configurations. These models can evaluate different design options rapidly and recommend modifications that enhance binding efficiency and specificity. Additionally, AI tools can predict the immunogenicity of BsAb candidates and simulate their manufacturing processes. By applying these AI technologies, we expedite the development of tailored BsAbs with higher therapeutic efficacy, reduced side effects, and improved patient outcomes.

Fig.1 Overview of antibody design AI agent.¹

Strategies for BsAb Structure Optimization by AI

AI-driven optimization of BsAb involves several strategies. First, ML models analyze sequence data to predict the optimal pairing of variable regions for enhanced antigen-binding specificity. DL models help refine structural predictions by simulating the 3D conformations of these antibodies, identifying potential stability issues, and suggesting modifications. Additionally, AI-based systems perform virtual screening to predict binding affinities with multiple targets simultaneously. In silico optimization also assesses the impact of mutations, stability, and manufacturability, providing a holistic approach to design. Combining these strategies accelerates the process of refining BsAb characteristics, ensuring higher efficacy, reduced immunogenicity, and improved therapeutic outcomes.

Published Data

The study presented a case where a BsAb showed high yield and routine solution appearance during discovery but went through substantial precipitation under agitation stress during 15 L CMC (Chemistry, Manufacturing, and Control) production. Using analytical tools, structural analysis, in silico predictions, and wet-lab validation, the molecular causes of precipitation were identified and resolved. Sequence engineering to reduce surface hydrophobicity and improve conformational stability effectively resolved agitation-induced aggregation. The optimized BsAb sequences enabled successful large-scale production in CMC. This case study highlights a potential protein engineering approach for evaluating and optimizing BsAbs with similar issues and underscores the importance of collaboration between structural modeling and CMC teams.

Fig.2 Structural modeling and stability design for BsAb optimization.²

Creative Biolabs provides customized AI-assisted BsAb optimization service, assisting precise design, enhanced binding, and rapid development.