Build an X-ray model in the browser
Automated main-chain tracing, sequence docking and side-chain building into the electron density, in cycles with maximum-likelihood refinement (Sextant Refine), all on the GPU. Learned density scorers find Cα positions, judge traced fragments and recognise residue types. Start from a molecular-replacement model and its sequence; finish by hand in Refinement mode and build again.
Build cycles
| Cycle | Residues | With sequence | Fragments | R-work | R-free |
|---|
Refinement
| Cycle | R-work | R-free | Gap | Bonds (Å) | Angles (°) | ⟨B⟩ (Ų) | Aniso | Waters | Clashes | wxyz | Time |
|---|---|---|---|---|---|---|---|---|---|---|---|
| No run yet. | |||||||||||
The MTZ keeps your observations and R-free flags and adds 2mFo−DFc (FWT/PHWT) and mFo−DFc (DELFWT/PHDELWT) map coefficients, FOM and Fmodel; Coot opens FWT and DELFWT automatically. The PDB carries ANISOU records for anisotropic atoms.
R-factors by cycle
R-work (solid) and R-free (dashed) after each macrocycle. R-free should fall with R-work; a widening gap means the model is fitting noise.
R-factors by resolution
Final model, in shells of equal reflection count. R rises towards high resolution as the signal weakens.
B-factors and the ADP model
Equivalent isotropic B of the final model: protein and ligand atoms (bars) and waters (outline). Anisotropic atoms are counted at their Beq.
Ordered solvent
| Step | Before | Added | Removed | After | R-free, just after |
|---|---|---|---|---|---|
| No run yet. | |||||
ΔLL is the change in −log-likelihood on the half of the data that did not see the peak (negative = the water explains those reflections better). Click a water to centre the 3D view on it.
Geometry and validation
Table 1
Data collection and refinement statistics in the usual journal layout. Cells marked fill in need the unmerged data; type the values from your processing log straight into the table before copying.
Electron density
Drag to rotate · shift-drag to pan · wheel to zoom · shift-wheel to change the 2mFo−DFc contour · double-click an atom to centre on it.
Model building
The agent reads the 2mFo−DFc and mFo−DFc maps residue by residue and tries what a crystallographer would: rotamers, incomplete side chains, peptide flips, main-chain zones, waters, obvious ions, missing residues and sequence mismatches. A change is proposed only when density and geometry both improve. Nothing enters the model until you accept it; then reciprocal-space refinement follows and the model is scanned again, with R-free as the referee (a round that raises R-free is undone).
| Round | Changes | R-work | R-free |
|---|
Ligands
Pick a ligand. L puts it at the centre of the view for placing by hand (Shift-drag moves it, Alt-drag turns it), F then fits it around that spot, and Find ligand sites tries every blob of unexplained density. The fit uses only the map and the restraints. Nothing joins the model until you press Enter.
Search log
Sequence
CC to 2mFo−DFcDifference density
Connected mFo−DFc density above +3σ after refinement, ranked by integrated density. Isolated blobs are candidate waters, ions or ligands; blobs on the model point to alternative conformations or misplaced atoms.