Related Experiment Video
Updated: Aug 8, 2026
12:30
Structural Information from Single-molecule FRET Experiments Using the Fast Nano-positioning System
Published on: February 9, 2017
Assessing Structural Prediction Accuracy for Nanobody-Small Molecule Complexes
Berta Bori-Bru1,2,3, Juan-Pablo Salvador2,4, Ramon Crehuet1
1Computational and Theoretical Chemistry Group (QTC), Institute for Advanced Chemistry of Catalonia (IQAC) - CSIC, Barcelona, Spain.
Protein Engineering, Design & Selection : PEDS
|August 7, 2026
Summary
Structure prediction models show promise for nanobody-small molecule design, accurately modeling nanobody structures but facing challenges in ligand positioning. CDR1 is key for binding, not CDR3.
Area of Science:
- Structural bioinformatics
- Computational drug discovery
- Protein engineering
Background:
- Generative models have advanced antibody and nanobody design.
- Nanobody engineering for small molecule sensing is an emerging field.
- Accurate structure prediction is crucial for designing effective nanobody biosensors.
Purpose of the Study:
- To evaluate state-of-the-art structure prediction tools for nanobody-small molecule complexes.
- To identify the strengths and limitations of these tools in predicting binding interactions.
- To guide future development of nanobody-based sensors.
Main Methods:
- Assessed seven structure predictors: AlphaFold3, Chai-1, Boltz-2×, RoseTTAFold3, Protenix, FlowDock, and OmegaFold.
- Analyzed nanobody and CDR (complementarity-determining region) geometries and ligand placement.
- Performed contact analysis to identify key binding residues.
- Evaluated the utility of intrinsic confidence scores for predicting binding accuracy.
Main Results:
- Most predictors accurately modeled nanobody and CDR structures but struggled with ligand orientation.
- Co-folding approaches improved the overall accuracy of predictions.
- Complementarity-determining region 1 (CDR1) was found to be primarily involved in ligand binding, contrary to expectations for CDR3.
- Intrinsic confidence scores showed poor correlation with experimental accuracy and limited ability to differentiate binders.
Conclusions:
- Current structure prediction methods show potential for nanobody-small molecule complex modeling but require refinement for ligand interaction prediction.
- CDR1 plays a significant role in small molecule recognition by nanobodies.
- Confidence scores from predictors are not reliable indicators of binding accuracy in this context.
