While many of therapeutic monoclonal antibody rely on their highly specific and high affinity binding to their targets, we have previously reported that Fab of antibodies can be engineered to have pH dependent, calcium ion dependent or ATP dependent antigen binding. We now report novel antibodies in which the same paratope of Fab can be engineered to bind to multiple antigens having very low homology. These antibodies are now being tested in phase 1 clinical study.

The Specifica Generation3 Library Platform is based on highly developable clinical scaffolds, into which natural CDRs purged of sequence liabilities have been embedded. The platform directly yields highly diverse, subnanomolar, developable, drug-like antibodies more potent than those from immune sources. This talk will discuss the extension of the platform to the direct selection of pH sensitive antibodies: binding better at pH 6.0, or binding better at pH 7.4.

The antibody repertoire generated by an animal in response to immunization results from its recognition of the target antigen, its native genetic diversification and cellular selection mechanisms, and the sequences of its immunoglobulin genes. All of these parameters are profoundly influenced by the host animal species and its genetics. OmniAb® accesses the biodiversity of six species to generate high-quality custom repertoires of human antibodies to empower therapeutic antibody discovery for a wide variety of targets and workflows.

Twist Biopharma Solutions (TBS), a division of Twist Bioscience, combines DNA synthesis with antibody engineering expertise to provide end-to-end antibody discovery solutions. The result is a make-test cycle engine that yields better antibodies against challenging targets utilizing immunization, libraries, and machine learning. TBS continues to expand its capabilities in partnership with others to further utilize their make-test cycle.

To gain insights into IgM’s assembly mechanics that underwrite their high-level secretion, we characterized the biosynthetic process of a natural human IgM using a HEK293 cell platform. By creating a series of mutant subunits that differentially disrupt secretion, folding, and specific inter-chain disulfide bond formation, we assessed their effects on various aspects of IgM biosynthesis. The mutations caused a spectrum of changes in steady-state subcellular subunit distribution, ER-associated inclusion body formation, intracellular subunit detergent solubility, covalent assembly, secreted IgM product quality, and secretion output. Through this combinatorial approach, we consolidated overlapping yet fragmented knowledge on IgM biosynthesis while unexpectedly revealing that the loss of certain inter-chain disulfide bonds was tolerated in polymeric IgM assembly and secretion. The findings demonstrate the crucial role of underlying non-covalent protein-protein interactions in orchestrating the initial subunit interactions and maintaining the polymeric IgM product integrity during ER quality control steps, secretory pathway trafficking, and secretion.

Biocytogen has developed a family of megabase-scale gene edited mice to expediate the generation of fully human antibody binders and TCR binders. Among them, RenLite is suitable for Common Lite Chain antibody discovery, and RenNano is specifically for human nanobody discovery. Half million high-quality antibodies for over one thousand human therapeutic targets is open for licensing and collaboration.

Determining the specificity of adaptive immune receptors- antibodies, and T cell receptors (TCRs) - is critical for understanding immune responses and advancing immunotherapy and drug discovery. Immune receptors exhibit extensive diversity in their variable domains enabling them to interact with a plethora of antigens. Despite the significant progress made by AI tools such as AlphaFold2 and AlphaFold3 in predicting protein structures, challenges remain in accurately modeling the structure and specificity of immune receptors, primarily due to the limited availability of high-quality crystal structures and the complexity of immune receptor-antigen interactions. Here, I will present advancements in sequence-based approaches for training machine learning models that predict immune receptor specificity.

Given the difficulties in discovering novel therapeutic antibodies, MOLCURE has created a platform that combines AI, laboratory automation, and molecular biology experiments. In this presentation, we will showcase the performance of our AI-generated antibodies, including pM-level Kd values and a variety of target epitopes. Furthermore, we will propose generative AI methods for designing antibodies with desired functionalities, which require minimal experimental validation.

Explore the essentials of collaborations between scientists and AI teams to understand the opportunities, challenges, and risks involved in AI-driven antibody design and how to best leverage data science and data scientists. Key topics include: project fit and feasibility using AI; real-world use cases of failure and success; optimal data to support AI-driven antibody design; communication challenges and opportunities between technologists and scientists; and data protection and intellectual property. Leave the presentation with a better understanding of how to leverage AI teams for your next antibody discovery and engineering campaign.

The swift identification of promising antibody candidates from various generation methods is crucial for driving therapeutic development. This presentation examines the practical role of in silico analysis in expediting this process. Utilizing adaptable and user-friendly bioinformatics tools, we demonstrate how streamlined pipelines improve efficiency, aid in result interpretation, and facilitate the selection of optimal candidates across experiments. In this talk we present how Chiome Bioscience effectively uses the PipeBio bioinformatics platform to support and accelerate antibody discovery pipelines at the company.

We introduce AbDiffuser, an equivariant and physics-informed diffusion model for the joint generation of antibody 3D structures and sequences. AbDiffuser is built on top of a new representation of protein structure, relies on a novel architecture for aligned proteins, and utilizes strong diffusion priors to improve the denoising process. Our approach improves protein diffusion by taking advantage of domain knowledge and physics-based constraints; handles sequence-length changes; and reduces memory complexity by an order of magnitude, enabling backbone and side chain generation. We validate AbDiffuser in silico and in vitro. Numerical experiments showcase the ability of AbDiffuser to generate antibodies that closely track the sequence and structural properties of a reference set. Laboratory experiments confirm that all 16 HER2 antibodies discovered were expressed at high levels and that 57.1% of the selected designs were tight binders.

Despite the central role that antibodies play in modern medicine, there is currently no way to rationally design novel antibodies to bind a specific epitope on a target. I will discuss the development of a deep-learning pipeline capable of designing de novo antibodies that bind to user-specified epitopes. This pipeline designs diverse antibodies against several types of epitopes, the designs are readily affinity-optimized and we demonstrate that, for one design, the pipeline achieves atomic-level accuracy versus a cryo-EM structure.

AI's potential to create antibodies from scratch is promising but hampered by poor hit rates and binding strengths, rooted in insufficient training data. We have addressed this issue by using computational simulations to determine data requirements such as modality, amount, and diversity. Simulations have been guiding our ongoing experimental data generation work, marking a shift towards a data-centric strategy that complements recent algorithmic progress, aiming to overcome current challenges.