This page explains how 3decision aligns two structures, and why you might choose one method over the other. For the options on screen, see Superposition.
3decision has two methods:
This method uses the pockets that 3decision detects on every structure at registration (see Pocket Browser). The reference pocket of the reference structure drives the alignment.
The method works when the two pockets overlap enough and have a similar arrangement in space. That makes it the natural choice for comparing binding sites of the same or closely related proteins, for example a series of ligand-bound structures of one target.
This method uses the sequence mapping done at registration, which links each chain to one or more biomolecules (UniProt entries or generic sequence identifiers).
You choose a chain on each structure, and optionally a residue range. 3decision aligns the two sequences and superposes the structures on the aligned residues. When the chains belong to different proteins, the residue range is taken from the sequence alignment.
The fit is best when both chains are the same protein, but close orthologs also align well. Because it does not depend on pockets, this method suits multi-chain complexes, antibodies, and comparisons of one domain.
| Your question | Method |
|---|---|
| How do these ligands bind the same site? | By pocket |
| How does the binding site differ between two related targets? | By pocket |
| How does an antibody–antigen complex differ from another one? | By chain |
| Has one domain moved relative to the rest of the protein? | By chain, with a residue range on the stable domain |
| The pockets are too different to align | By chain |
Whatever the method, the fit is optimized on the aligned region only. Parts far from that region can look shifted simply because they were not used in the fit. Keep this in mind when you read Highlight Mode results outside the aligned region.