Antibody R&D Collaboration

Compared with small molecule drugs, antibody drugs have shown rapid growth in the field of drug R&D in recent years due to their strong binding specificity, small side effects, high success rate, and ability to target protein-protein interaction. CADD+AI prediction method can assist the development of antibody drugs, significantly speed up the R&D process, improve the success rate, and reduce the cost and risk.

Our Scope of Work

  • Murine antibodies humanization
  • Antibody immunogenicity (ADA) elimination
  • Prediction and improvement of antibody developability
  • B cell epitope prediction
  • T cell epitope prediction
  • Co-development of other customized AI models

Value Added

  • Antibody humanization technology reduces immune response from murine antibodies
  • Epitope prediction guides the immunogenicity (ADA) elimination of antibodies
  • Assess, predict, and optimize the developability of antibody
  • B cell epitope prediction helps to avoid epitope patent protection
  • T cell epitope prediction model can accurately locate neoantigen
Murine antibody humanization

Murine antibodies produced by hybridoma technology tend to be highly immunogenic. Humanization methods such as CDR grafting, surface remodeling, and chain replacement are used to eliminate immunogenicity. Further combining the Al models such as T cell and B cell epitope prediction, we can reduce anti-drug antibody (ADA) response and improve the success rate of clinical trials.

Antibody developability prediction

The antibody’s aggregation, viscosity, solubility, immunogenicity, stability, expression yield, post-translational modification site, are important characteristics closely linked to its overall developability. A less-than-ideal developability can greatly affect the subsequent CMC process and clinical trials. There are no comprehensive and complete solutions to predict and improve the developability of antibodies currently. Based on the features extracted from protein sequence and structure, AI models can accurately predict various antibody’s developability-related properties, help scientists improve on the unsatisfactory properties, and lower the overall risk of clinical failure.

Antibody de novo design

Obtaining antibodies that bind to a given epitope can circumvent the patent protection of epitopes. Traditional method to generate antibodies is to use linear peptides from antigenic proteins as antigens, which may lead to problems such as low antibody binding affinity and non-specific binding, which can be avoided by antibody de novo design.

Tumor neoantigen prediction

In the cell therapy of tumor immunity, how to identify tumor neoantigens with high specificity and strong immunogenicity is a crucial step. It is highly costly to conduct neoantigen screening through wet-lab experiments. The big-data-driven Al model can accurately predict tumor neoantigens with high credibility and provide personalized medical guidance for cell therapy.

Core Technology

Artificial Intelligence Cloud Computing