Hu-mAb

Hu-mAb predicts and suggests mutations to humanize monoclonal antibody variable domain sequences to reduce immunogenicity while maintaining therapeutic efficacy.

Key Features:

• Machine Learning Approach: Hu-mAb uses machine learning classifiers trained on extensive antibody repertoire sequence data to distinguish human from non-human variable domain sequences and predict immunogenicity with reported performance exceeding existing models.
• Data-Driven Humanization: By analyzing large-scale sequence data, Hu-mAb identifies specific mutations to introduce into an input antibody sequence to reduce predicted immunogenicity based on a negative correlation between classifier output scores and experimental immunogenicity of therapeutics.
• Efficiency and Accuracy: The mutations proposed by Hu-mAb closely align with mutations determined experimentally for known therapeutic antibodies, providing a computational alternative to trial-and-error humanization methods.

Scientific Applications:

• Immunotherapy: Humanization of monoclonal antibodies to reduce patient immune responses and enable safer therapeutic antibodies.
• Drug Development: Guiding sequence modification in antibody engineering to accelerate development of antibody therapeutics with lower immunogenicity.

Methodology:

Machine learning classifiers trained on extensive antibody repertoire sequence data analyze large-scale sequence data to propose mutations for input variable domain sequences; classifier output scores are observed to correlate negatively with experimental immunogenicity of therapeutics.