Dynamic PDB: A New Dataset and a SE(3) Model Extension by Integrating Dynamic Behaviors and Physical Properties in Protein Structures

Dynamic PDB contains 12.6K proteins subjected to 1 microsecond of molecular dynamics simulations with detailed physical properties such as atomic velocities, forces, and potential/kinetic energies
Select high-resolution structures (≤ 2.0 Å) proteins with fewer than 500 AA, excluding membrane proteins and proteins with coil-loop conformations
Use OpenMM (Amber-ff14SB force field) for MD simulation with two equilibration phases: a canonical ensemble (NVT) and an isothermal-isobaric ensemble (NPT)
Build SE(3)-diffusion with conditional physical properties (velocity and force) and Invariant Point Attention