bioRxivJClub
Membrane Geometry Reveals Strengths and Limitations of AlphaFold-Predicted Transmembrane Protein Complexes
bioRxiv · · Preprint · Open access
Sziklai, Wiener + 2 more
Abstract ↗AI summary
The abstract is read at the publisher; the summary is JClub's.
AlphaFold-Multimer predicts 189,469 high-confidence transmembrane protein complexes, with 95.4% of 125,657 evaluable multi-helix TM-TM cases matching expected 15–35 Å hydrophobic thickness.
- Why it matters: Confidence scores alone can miss assemblies that violate lipid-bilayer constraints, especially for transmembrane interactions where geometry and membrane placement restrict plausible complexes.
- What they did: The authors evaluated membrane compatibility across AlphaFold Protein Structure Database predictions, then tested whether AlphaFold3 (AF3) improves assembly validity using SERCA-phospholamban and norepinephrine transporter cases plus a 173-target cholesterol benchmark.
- The result: Even with favourable confidence, some complexes show membrane-position or protomer-orientation inconsistencies, AF3 recovered SERCA-phospholamban but not the cholesterol-mediated norepinephrine transporter homodimer, and cholesterol-binding poses were limited, especially for targets released after the AF3 training cutoff.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.
- Open access