Complexes

Aug 25, 2026

Antigen size biases AlphaFold3 antibody-antigen confidence

Aug 25, 2026

ipTM is sensitive to construct length even when predicted interfaces are unchanged

Jul 28, 2026

Coevolutionary patterns in multiple sequence alignments do not contribute to protein-protein complex prediction

Jul 28, 2026

MSA protein language models can identify intermolecular contacts

Jul 28, 2026

Structure prediction and design tools trained on monomers generalize to oligomers

Jul 17, 2026

Antibody-antigen complex prediction scales with training data and compute

Jul 17, 2026

Antibody-antigen modeling by diffusion-based structure prediction is data-limited

Jul 17, 2026

Different AlphaFold3 clones have differently calibrated confidence heads

Jul 17, 2026

Interface contact score

Apr 21, 2026

AlphaFold2 recapitulates interaction biases from PDB when modeling antibodies

Apr 21, 2026

Antigen context improves CDRH3 structure prediction

Apr 21, 2026

Correct CDRH3 prediction is necessary but insufficient for correct Ab-Ag docking

Apr 21, 2026

Correct antibody-antigen prediction in AF3 and related models is partially determined by training set similarity

Apr 21, 2026

Dropout improves AF2 multimer prediction, including for antibody-antigen complexes

Apr 21, 2026

Increasing diffusion samples is sufficient to yield correctly predicted antibody-antigen complexes

Apr 21, 2026

The structures of nanobodies with kinked CDR3 loops are more difficult to predict in complex with their antigen

Apr 21, 2026

AF3-generation methods can be substantially shrunk without sacrificing prediction quality, except on PPIs

Apr 21, 2026

All-atom protein structure prediction methods are unable to model PPIs with large buried surface areas

Apr 21, 2026

AlphaFold3 learns intramolecular interactions faster than intermolecular interactions

Apr 21, 2026

Correct protomer structure prediction is necessary but insufficient for accurate protein-protein docking

prediction/complexes

Apr 21, 2026

DockQ

Apr 21, 2026

Kinks in nanobody CDR3 loops are correctly predicted by structure prediction methods

Apr 21, 2026

Protein ensemble prediction methods do not generate conformations that are amenable to PPI docking

Apr 21, 2026

Protein-ligand co-folding

Apr 21, 2026

Structure prediction metrics can be used to predict oligomer stoichiometry

Apr 21, 2026

Structure prediction neural networks predict different conformations of a given sequence depending on if a single chain is predicted versus multiple chains

Apr 21, 2026

All-atom structure prediction of RNA is driven by memorization

Apr 20, 2026

AlphaFold3

Apr 20, 2026

Quartz 4

Explorer

Complexes

Antigen size biases AlphaFold3 antibody-antigen confidence

ipTM is sensitive to construct length even when predicted interfaces are unchanged

Coevolutionary patterns in multiple sequence alignments do not contribute to protein-protein complex prediction

MSA protein language models can identify intermolecular contacts

Structure prediction and design tools trained on monomers generalize to oligomers

Antibody-antigen complex prediction scales with training data and compute

Antibody-antigen modeling by diffusion-based structure prediction is data-limited

Different AlphaFold3 clones have differently calibrated confidence heads

Interface contact score

AlphaFold2 recapitulates interaction biases from PDB when modeling antibodies

Antigen context improves CDRH3 structure prediction

Correct CDRH3 prediction is necessary but insufficient for correct Ab-Ag docking

Correct antibody-antigen prediction in AF3 and related models is partially determined by training set similarity

Dropout improves AF2 multimer prediction, including for antibody-antigen complexes

Increasing diffusion samples is sufficient to yield correctly predicted antibody-antigen complexes

The structures of nanobodies with kinked CDR3 loops are more difficult to predict in complex with their antigen

AF3-generation methods can be substantially shrunk without sacrificing prediction quality, except on PPIs

All-atom protein structure prediction methods are unable to model PPIs with large buried surface areas

AlphaFold3 learns intramolecular interactions faster than intermolecular interactions

Correct protomer structure prediction is necessary but insufficient for accurate protein-protein docking

DockQ

Kinks in nanobody CDR3 loops are correctly predicted by structure prediction methods

Protein ensemble prediction methods do not generate conformations that are amenable to PPI docking

Protein-ligand co-folding

Structure prediction metrics can be used to predict oligomer stoichiometry

Structure prediction neural networks predict different conformations of a given sequence depending on if a single chain is predicted versus multiple chains

All-atom structure prediction of RNA is driven by memorization

AlphaFold3

Antibody structure prediction