When I run Protein Structure Alignment in Maestro or structalign from the command line on pairs of proteins separately I get slightly different alignments compared to when I run on the full set of proteins all together. For example,

$SCHRODINGER/utilities/structalign <pdbid1>.pdb <pdbid2>.pdb <pdbid3>.pdb

gives a slightly different alignment than

$SCHRODINGER/utilities/structalign <pdbid1>.pdb <pdbid2>.pdb
$SCHRODINGER/utilities/structalign <pdbid1>.pdb <pdbid3>.pdb.

Is this a bug?

No, these are actually different calculations. When structalign (SKA) does a multiple alignment it calculates a matrix of pairwise scores for all pairs of structures. The structures are added to the alignment starting with the most similar pair, and sequentially adding the next-most-similar to the average coordinates of the already-aligned structures.

Note, when the proteins are aligned together, the frame of reference for the alignment is the coordinate frame of the superposition (i.e. the query is never rotated), but this doesn't mean that all alignments are pairwise alignments to the query.

Back to Search Results