CAMEO
TrustBCAMEO's rolling evaluation solves CASP's staleness problem but introduces its own biases. The target set is whatever lands in PDB that week, which skews toward crystallography-friendly proteins. Still, it's the best continuous signal we have for tracking server performance over time.
What It Measures
Continuous, automated evaluation of protein structure prediction servers. Runs weekly against newly released PDB structures, providing a rolling leaderboard rather than a point-in-time competition.
What It Doesn't Measure
Non-server methods, local installations, or any model that can't be run as an automated prediction service. Also doesn't evaluate drug-discovery-relevant metrics.
Maintainer
SIB Swiss Institute of Bioinformatics
https://www.cameo3d.org/ →Known Limitations
Only evaluates methods deployed as automated servers — excludes important models that aren't set up as web services. Weekly target set is small and biased by PDB deposition patterns. Doesn't capture the full diversity of structurally interesting problems.