Information on beta server
Last updated 08-04-05The beta server is a complete reimplementation of most of the original Morph Server, extended from the old multichain server. Unlike the original, it will handle nucleic acids and arbitrarily large complexes. Unlike the old multichain server, it will also handle non-homogeneous input structures. Both traditional adiabatic mapping and FRODA interpolations are available.
This server is being built both as an animation tool and a method of analyzing multi-subunit complexes; it does not, however, produce the statistics provided by the original server. Although it accepts homologous structures, it does not employ the same type of homogenization strategy originally used. Output structures are more faithful to the originals, but won't always look as pretty.
In brief, the strategy for processing PDB files is as follows:
- The requested chains are extracted from each PDB file; non-standard amino acids (selenomethionine, phosphorylated residues) will be converted to their equivalent generic residue. Gaps are detected and treated as unknown residues in the sequence.
- Any chains that are not identical are aligned using ClustalW.
- The alignment is scanned to determine a consensus sequence. The residue numbering of the first structure will not change; the numbering of the second structure will be modified to fit.
- Gaps will be removed or introduced where appropriate, but no atoms will be filled in.
- Currently, mutations are reduced to alanine.
- The structures are fed into CNS, which generates a topology and eliminates extra atoms. The output files are identical except for coordinates.
A few limitations currently remain:
- Aligned residues that are not the same in both structures are mutated to alanine (or glycine where appropriate). This means that all atoms in the output structure represent atoms in the PDB file. However, it also results in skeletonized structures where homology is distant.
- The list of chains to morph must be identical for each file. You must ensure that chain IDs are valid across both structures. (This is not necessarily the case in official PDB structures, unfortunately.)
- Sieve-fitting is not done. However, unless otherwise specified, the server will perform an LSQ-fit using CNS. If you disable this, the structures need to be already in the proper orientation.
PDB files will often fail to meet the standards required for this to work. However, the changes necessary are less drastic than in previous versions. No manual modification of sequence or structure should be mandatory, but chain IDs will often need to be changed, easily done using PyMOL or an equivalent program. You may, if you want, alter your structures further if you wish to work around some of the limitations of this server.
Heteroatoms are currently disabled, but will be reincorporated soon. Users are warned that heteroatom processing is usually nightmarish, and different PDB files rarely use the same numbering.
--Nat Echols
Copyright 1995-2005
Mark.Gerstein@yale.edu
Last modified 8-21-01