CG → AA Back-mapping
How atomistic coordinates are reconstructed from coarse-grained trajectories for downstream inspection and analysis.
Interpretation note. The back-mapped coordinates are intended for inspection, comparative structural analysis, and contact-based measurements. They should be read as a reconstruction workflow derived from the CG trajectory, not as a substitute for an independently relaxed atomistic simulation.
▶ THE COOPERATIVE PRINCIPLE — AA positions inherit the propagated simulation state
CG Forces
Classical dynamics
Drive the base motion of all beads.
QCloud Corrections
Selective refinement inputs
Refine forces on prioritized regions.
ML Residual
Learned patterns
Fill in corrections between QCloud evaluations.
four-step method
▶ METHOD — click each step to see the math
1. Bead-to-Atom Mapping
Labels → residue ranges
Each CG bead label encodes chain + residue range. Resolves to all constituent AA atoms from reference PDB.
2. Interpolation Weights
Distance-weighted, same-chain
wi = (1/di) / ∑(1/dj)
Nearest 2 beads on same chain. Snap to weight 1.0 within 0.01Å.3. Displacement Interpolation
Apply weighted CG deltas
rAA(a)(t) = rAA(a)(0) + ∑i wi [rCG(i)(t) − rCG(i)(0)]
At t=0, back-mapped = original crystal structure (Δ=0).4. Bond Relaxation
8 iterations of constraint fixing
C-N target 1.33Å (45% push/pull). CA-CA target 3.80Å (30% pull). Entire residues dragged to preserve geometry.
output files
▶ OUTPUT FILES — all PDB files load in Chimera, VMD, or PyMOL
aa_backmapped_trajectory.pdb
Multi-model AA trajectory
aa_backmapped_initial/final.pdb
First and last AA frames
aa_backmapped_combined.pdb
Initial + final in one file
aa_reference.pdb
Original crystal structure