A complete, multi-level conformational clustering of antibody complementarity-determining regions

Area of Science:

• Biochemistry
• Structural Biology
• Immunology

Background:

• Antibody complementarity-determining region (CDR) conformation classification is crucial for antibody modeling, engineering, and sequence-to-structure relation inference.
• Previous studies often used reduced structure sets or applied pre-filtering, potentially limiting comprehensive analysis.

Purpose of the Study:

• To create a complete and non-redundant repertoire of all available CDR conformations.
• To improve future structure validity assessments and antibody engineering strategies.

Main Methods:

• Employed a multi-level statistical clustering approach.
• Utilized structure superposition Root-Mean-Square Deviation (RMSD) for first-level clustering.
• Incorporated Ramachandran regions for second- and third-level qualitative classification.

Main Results:

• Classified a total of 12,712 CDR conformations.
• Generated a rich CDR Knowledge-Base with detailed annotations and cluster descriptions.
• Presented a novel nesting of conformational sensitivity and specificity.

Conclusions:

• The developed CDR repertoire offers an improved understanding of antibody conformational space.
• Provides a systematic framework for enhanced sequence-based prediction and knowledge-based antibody engineering, including humanization.