Computational Antibody Design

Antibody calculation design

In recent years, as the research and development of China's biological drugs, especially antibody drugs, has become more and more popular, the application of computational simulation technology to innovative biological drugs has also become a means for major companies to gradually rise. Computational Antibody Design is increasingly used in the design of antibodies against different targets. The current common computational antibody design method starts with the modeling of antibodies and antibody-antigen complexes. Select antibody sequence for antigen binding experiment test, for example, bind to soluble antigen or cell-expressed antigen, and further optimize the binding substance through affinity maturation method. However, this type of method often requires modeling of the overall structure, which requires a large amount of calculation and takes a long time. Therefore, an antibody design based on the CDR of the specific binding site of the antibody and the antigen was proposed. This method simplifies the calculation and design of the antibody to the modeling and design of the antibody CDR, which greatly reduces the calculation requirements.

Computational antibody design platform

At present, the commonly used computational antibody design platforms are mainly Discovery Studio and Rosetta developed from protein structure simulation, and Schrdinger developed from chemical drug structure simulation.

Computational antibody design platform

Evaluation of computational antibody design

The therapeutic monoclonal antibody must not only bind to the target, but also must not have developability problems such as poor stability or high aggregation. Although the discovery of small molecule drugs benefited from Lipinsky's five principles to guide the selection of molecules with appropriate biophysical properties, there are currently no guidelines for antibody design. In 2019, Matthew IJ Raybould and others published a study on PNAS, modeling the structure of a large number of antibody variable domains in phase 1 clinical phase antibody therapies (CSTs), and summed up five indicators for evaluating antibody development: CDR total Length, range and size of surface hydrophobicity, positive and negative charge of CDR, asymmetric charge on the surface of heavy and light chains.

Evaluation of computational antibody design

The computational antibody design platform usually includes the following functions:

  1. Annotation of antibody sequence: annotation of antibody heavy and light chain variable regions, constant regions, and CDR regions, and recognition of antibody Gemline genes;
  2. Analysis of antibody sequence: prediction of antibody post-translational modification sites (such as glycosylation sites, etc.), and antigen linear epitope prediction;
  3. Prediction and optimization of antibody structure, prediction of full-length antibody, antibody Fab region, and scFv structure;
  4. Antibody-antigen interaction prediction, antigen conformation epitope recognition;
  5. Design and modification of antibody molecules: antibody affinity maturation, antibody stability optimization, bispecific antibody design, antibody aggregation effect analysis, antibody humanization modification.


  1. Raybould M, Marks C, Krawczyk K, et al. Five computational developability guidelines for therapeutic antibody profiling. Proceedings of the National Academy of Sciences, 2019.
* For Research Use Only.




© 2021 CD ComputaBio. All rights reserved.