How do pLDDT and PAE answer different structure-confidence questions?

Why

Imagine a protein with two well-folded domains connected by a flexible linker. Each domain might have very high pLDDT because its internal fold is confidently predicted. Yet PAE between the domains may be high, indicating that their relative orientation is poorly constrained.

Use pLDDT to judge local secondary structure and candidate site geometry. Use PAE to evaluate domain placement, hinge arrangements and predicted interactions across regions. Neither score establishes ligand binding, physiological oligomeric state or molecular dynamics.

When communicating a model, show both the 3D coloring and PAE matrix instead of reporting one average number. For a proposed inter-domain pocket, high PAE makes a precise docking pose uncertain even if surrounding residues have high pLDDT.