Antibody design using LSTM based deep generative model from phage display library for affinity maturation.

Area of Science:

• Biotechnology
• Immunology
• Computational Biology

Background:

• Therapeutic antibody development relies on molecular evolution and affinity maturation.
• Current affinity maturation techniques are costly and time-consuming due to extensive mutation experiments.
• Exploring antibody sequence space efficiently is crucial for discovering high-affinity antibodies.

Purpose of the Study:

• To develop an efficient deep learning-based method for antibody sequence generation and prioritization.
• To apply this method for the affinity maturation of antibodies targeting kynurenine.
• To demonstrate the superiority of the machine learning approach over traditional screening methods.

Main Methods:

• Utilized a Long Short-Term Memory (LSTM) network, a deep generative model, for sequence generation.
• Employed phage display panning with a kynurenine-binding oriented human synthetic Fab library.
• Trained the LSTM model using next-generation sequencing (NGS) data of binding antibody sequences.
• Correlated the likelihood of generated sequences with their binding affinity.

Main Results:

• The trained LSTM model effectively generated antibody sequences with high binding affinity.
• Generated antibody sequences exhibited over 1800-fold higher affinity compared to the parental clone.
• The machine learning approach outperformed frequency-based screening in identifying high-affinity sequences.

Conclusions:

• Deep generative models like LSTM can significantly enhance the efficiency of antibody affinity maturation.
• This computational approach offers a cost-effective and faster alternative to traditional experimental methods.
• The developed method holds promise for accelerating the discovery of novel therapeutic antibodies.