We use cookies to enhance your browsing experience and analyze site traffic. By continuing to use this site, you agree to our cookie policy. Privacy Policy
Based on AI structure prediction and energy modeling, we analyze residue spatial synergy to identify cooperative mutation sites. Through a 1+1>2 strategy, we improve affinity in small-scale libraries. Fewer experimental variants, higher screening accuracy, helping researchers obtain candidates with greater therapeutic potential at an early stage.
Validated Results
Why Is It Needed?
"In the natural immune system, affinity maturation refers to the process where B cells undergo somatic hypermutation and clonal selection in germinal centers, enabling antibodies to evolve stronger target-binding capabilities. However, animal immunization or conventional design alone can rarely achieve the levels required for drug development."
Key Advantages
Relies on random mutagenesis or saturation scanning with wide range but low precision; insufficient sampling introduces structural prediction errors.
Inefficient screening leads to large experimental volume, low success rate, and significantly increased time and costs.
Fails to consider cooperative effects of spatially proximal mutation sites, missing 1+1>2 synergy opportunities.
Combines protein language models and inverse folding models to analyze mutation effects on antigen-antibody binding from both sequence evolution and structural stability, improving mutation screening accuracy and intelligently recommending beneficial mutations.
Based on AI-predicted antigen-antibody complex structures, automatically analyzes residue spatial synergy, identifies cooperative mutation sites, and through a 1+1>2 strategy, helps users improve affinity in small-scale libraries.
Validation
The platform validates affinity maturation in real projects. Below are experimental comparisons for two library types — light chain sub-library and heavy-light chain sub-library — covering affinity, kinetics and developability metrics.
Among the 32 matured antibodies, the vast majority improved affinity by over 3-fold versus the parent (C48, KD 1.11 nM). The best antibody (red marker) lowered KD to 0.091 nM with a ~12-fold slower dissociation rate, while 31/32 samples expressed at higher total yield than the parent (median 1.41x) and most candidates maintained SEC-HPLC purity above 90% — balancing activity, kinetics and developability.
Pipeline
The platform supports two starting inputs: when an antigen-antibody complex structure is already available, analysis begins directly from the experimental structure; when only antigen and antibody sequences are available, AI structure prediction is performed first before downstream mutation design.
Path A
If the user already provides an antigen-antibody complex structure, the experimentally resolved structure is used directly as the starting point for mutation analysis without re-prediction.
Path B
If only antigen and antibody sequences are provided, the platform first predicts the antigen-antibody complex structure with AI before proceeding to mutation analysis.
Report
Below are partial screenshots from the affinity maturation service delivery report, showing the complete workflow from mutation design to affinity prediction evaluation.
Use Cases
Best suited for projects with an existing binding antibody that need higher affinity while preserving the basic binding mode.
Random mutagenesis covers broad space but has low hit rates and many low-value screens. The platform combines protein language models, inverse folding models, and structural information to prioritize mutations more likely to improve binding with lower structural risk.
Affinity improvement often does not come from a single site alone. Spatially proximal residues can have synergistic 1+1>2 effects. Structure-driven analysis identifies these combination opportunities so smaller sub-libraries have a better chance of meaningful improvement.
Fewer experiments, higher accuracy — helps you rapidly advance from candidate antibodies to high-affinity drug molecules.