Predicting Immunogenicity with AI and ML Tools
Advancing Biotherapeutic Design through Predictive and Generative Models
5/11/2027 - May 12, 2027 ALL TIMES EDT
As biotherapeutics become increasingly complex, the ability to predict and mitigate immunogenicity remains critical to successful drug development. Advances in AI, machine learning, deep learning, and immune-system modeling are enabling more integrated assessments of molecular, mechanistic, and clinical risk factors. Emerging approaches combine T cell and B cell epitope prediction, structural analysis, clinical datasets, and generative protein design to identify liabilities earlier and engineer safer, more developable molecules. Join leading industry and academic experts at Cambridge Healthtech Institute’s 3rd Annual Predicting Immunogenicity with AI and ML Tools conference to explore the latest computational strategies for improving risk prediction, informing candidate selection, and designing biotherapeutics with reduced immunogenicity.
Preliminary Agenda

PLENARY KEYNOTE SESSION

Panel Moderator:

PANEL DISCUSSION: From Mechanism to Modality: Smarter Drug Design in Autoimmunity

Yang Shen, PhD, Executive Director, Antibody Engineering, Bispecifics, Regeneron , Vice President , Bispecifics & Antibody Engineering , Regeneron

Panelists:

Stephen J. Demarest, PhD, CSO, Tentarix Biotherapeutics , CSO , Tentarix

Ryan P. Larson, PhD, Vice President & Head, Translational Sciences, Umoja Biopharma , SVP & Head , Translational Sciences , Umoja Biopharma

Kira Rubtsova, PhD, Associate Vice President, Eli Lilly and Company , Dir , Eli Lilly & Co

Qizhi Tang, PhD, Professor in Residence, Surgery, University of California San Francisco , Prof in Residence , Surgery , Univ of California San Francisco

IMMUNOGENICITY RISK ASSESSMENT

AI and ML NAMs in the Immunogenicity Risk Assessment and Mitigation of Proteins used in Therapeutic Applications

Photo of Zuben Sauna, PhD, Director, Division of Hemostasis, CBER, FDA , Director , Division of Hemostasis , FDA CBER
Zuben Sauna, PhD, Director, Division of Hemostasis, CBER, FDA , Director , Division of Hemostasis , FDA CBER

Immunogenicity of therapeutic proteins carries significant clinical and economic costs. This presentation examines how AI and ML-based New Approach Methodologies (NAMs) including can be used in conjunction with in vitro and ex vivo data to transform immunogenicity risk assessment and mitigation. Examples of integrated NAM workflows, establishing concordance with clinical outcomes and the use of mathematical models of immunogenicity for population-level and personalized immunogenicity prediction will be provided.

Beyond the Sequence: Using AI and ML to Integrate Molecular, Product, and Patient Data for Understanding Biotherapeutic Immunogenicity

Photo of Siddarth Prabhu, Process Development Scientist, Attribute Sciences, Amgen , Process Development Scientist , Attribute Sciences , Amgen
Siddarth Prabhu, Process Development Scientist, Attribute Sciences, Amgen , Process Development Scientist , Attribute Sciences , Amgen

RISK ASSESSMENT DATASETS

Presentation to be Announced

Building Standardized, High-Quality Preclinical and Clinical Datasets for Interpretable, Multivariable Risk Models

Photo of Timothy Hickling, PhD, Consultant, Quasor Ltd; CSO, Open Immune , Independent Immunogenicity Expert , Quasor
Timothy Hickling, PhD, Consultant, Quasor Ltd; CSO, Open Immune , Independent Immunogenicity Expert , Quasor

PROPERTY PREDICTION

Predicting Donor-Specific T Cell Recognition of Peptides for Biotherapeutics Immunogenicity Assessment

Photo of Mojtaba Haghighatlari, PhD, Senior Principal Machine Learning Scientist, Pfizer Inc. , Senior Principal ML Scientist , Pfizer
Mojtaba Haghighatlari, PhD, Senior Principal Machine Learning Scientist, Pfizer Inc. , Senior Principal ML Scientist , Pfizer

Immunogenicity risk assessment for biotherapeutics is well developed for peptide binding to and presentation by HLA class II (HLAII) molecules, but an immune response also requires that a presented peptide be recognized by a T cell receptor (TCR). Modeling this recognition is relatively limited by data availability. We present a workflow leveraging our internal T cell assay data, in which peptides from several candidate biologics are tested against donor samples with known HLAII allotypes, to train a model predicting donor-specific recognition. We integrate this model into the broader immunogenicity assessment workflow and discuss its potential to flag epitopes that pass both HLA presentation and T cell recognition filters.

Multimodal Foundation Models for Immunogenicity Prediction and Generation

Photo of Smita Krishnaswamy, PhD, Professor, Genetics & Computer Science, Yale University , Professor , Genetics & Computer Science , Yale Univ
Smita Krishnaswamy, PhD, Professor, Genetics & Computer Science, Yale University , Professor , Genetics & Computer Science , Yale Univ

Identifying immunogenic epitopes remains a bottleneck for cancer and infectious-disease vaccines. I will describe ImmunoStruct, which integrates peptide–MHC sequence, structure and biochemical properties across ~27,000 complexes to predict class-I immunogenicity, and our ongoing work on ImmunoFoundation, a self-supervised multimodal foundation model of TCR–peptide–MHC complexes coupling an ESM-2 sequence encoder with a graph transformer over predicted structure. ImmunoFoundation will focus on tasks such as generating immunogenicity-optimized novel TCRs.

RISK ASSESSMENT OF CELL AND GENE THERAPIES

The Many Faces of AAV Immunity: What do we Need to Measure?

Photo of Boris Gorovits, PhD, Vice President, Non-Clinical and Clinical Bioanalysis, Regeneron , VP Non-clinical and Clinical Bioanalysis , Regeneron
Boris Gorovits, PhD, Vice President, Non-Clinical and Clinical Bioanalysis, Regeneron , VP Non-clinical and Clinical Bioanalysis , Regeneron

AAV gene therapy triggers a multifaceted immune response that has potential to impact treatment safety and efficacy. Pre-existing and treatment-induced anti-capsid antibodies, whether neutralizing or binding, can block transduction, accelerate vector clearance, or drive complement-mediated reactions. Cellular responses against capsid or transgene product may eliminate transduced cells and cause loss of expression. Which responses to measure depends on route of administration, immune privilege of the target tissue, and transgene localization. This presentation will discuss these and related questions.

RISK MITIGATION

From TCR Sequencing to Immunogenicity Mitigation: Novel Approaches for Biopharmaceuticals

Photo of Bernard Maillere, PhD, Research Director, Immunology, CEA , Dir Research & Head , Cellular immunology , CEA/ University of Paris Saclay
Bernard Maillere, PhD, Research Director, Immunology, CEA , Dir Research & Head , Cellular immunology , CEA/ University of Paris Saclay

Understanding and mitigating T cell responses to biopharmaceuticals remain major challenges in therapeutic development. This presentation will highlight novel approaches to characterize biopharmaceutical-specific T cell responses through TCR sequencing, providing new insights into T cell response in ADA+ and ADA- patients, and T cell specificity using Tcr transfer into immortalized cell lines. It will also explore innovative protein engineering strategies to reduce immunogenicity while preserving therapeutic activity.

Protein Deimmunization with a Generative Diffusion Model

Photo of Kaiyi Jiang, PhD, Assistant Professor, Bioengineering, Princeton University , Assistant Professor , Bioengineering , Princeton Univ
Kaiyi Jiang, PhD, Assistant Professor, Bioengineering, Princeton University , Assistant Professor , Bioengineering , Princeton Univ

Anti-drug antibodies against protein therapeutics accelerate clearance, abolish efficacy, and can cause anaphylaxis. To reduce protein immunogenicity, we developed MESSI, a generative discrete diffusion model that redesigns sequences from structures while eliminating predicted MHC class II epitopes. MESSI achieves high structural recovery and sequences down to zero predicted epitopes. Deimmunized luciferase, a de novo PD-1 minibinder, and uricase designs exceeded parental activity, enabling safer and more effective protein therapeutics.


For more details on the conference, please contact:
Gemma Smith
Senior Conference Director
Cambridge Healthtech Institute
Phone: (+44) 7866 506 196

For sponsorship information, please contact:

Companies A-K
Jason Gerardi
Sr. Manager, Business Development
Cambridge Healthtech Institute
Phone: (+1) 781-972-5452
Email: [email protected]

Companies L-Z
Ashley Parsons
Manager, Business Development
Cambridge Healthtech Institute
Phone: (+1) 781-972-1340
Email: [email protected]


Register

View By:


Premier Sponsors

Biointron_new       Kactus