For once, I actually agree with Ian! I too refer to the process as
"molecular replacement", even if you don't run a "molecular replacement
program". For those who are interested in more than one opinion, other
popular "method used to determine the structure" in the PDB that I still
call "MR" are:
2639 fourier synthesis
447 difference fourier
96 rigid body refinement
39 rigid body
37 difference fourier plus refinement
31 isomorphous
31 difference fourier methods
25 refinement
21 rigid-body refinement
17 direct replacement
15 refmac
15 molecular substitution
15 molecular replacemen
15 difference map
14 native model phases
14 molrep
13 isomorphous molecular replacement
11 direct refinement
10 isomorphous with pdb entry xxxx
10 fourier difference
9 known phases
8 refined directly
8 isostructural to xxxx
8 difference fourier method
7 phaser
7 isomorphous to native
7 difference fourier from previously determined, related structure
6 rigid body refinement into new data
5 used previously- determined structure
5 direct based on known model
5 difference fourier maps
4 starting model was used without molecular replacement
4 protein structure is known in this cell
-James Holton
MAD Scientist
Ian Tickle wrote:
I would say it should still be classed as MR: the distinguishing
feature of MR is surely that it uses an known structure as a starting
model, not that it does a rotation/translation search. In any case
the distinction between a rigid-body search and RB refinement is moot.
At Astex our automated scripts routinely do a limited MR search for
all our protein-ligand structures, even though it may not be necessary
in all cases. This is because we have found from experience that RB
refinement alone is often not sufficient: presumably the degree of
non-isomorphism induced by soaking and/or freezing the crystals (e.g.
we often see 5-10% changes in cell parameters) takes it beyond the
radius of convergence of the refinement. We use a limited search MR
to avoid issues of finding symmetry-related or origin-shifted
solutions.
Cheers
-- Ian
On Mon, Apr 19, 2010 at 3:06 PM, Edward A. Berry <[email protected]> wrote:
Quite a lot of structures in the PDB involve minor variations on structures
that have already been solved (different ligands, mutants, high res), so the
"solution" involves refining the previous structure against the new data,
perhaps starting with rigid body refinement to correct minor variations
in cell parameters.
Would you include this in "molecular replacement"?
Ed
Nicholas Keep wrote:
Thanks to several people for helpful comments to my question on the
Proportion of MR in the PDB.
I got two very detailed responses one from the OCA team at the Weizmann
which went to the Bulletin Board
"This is what OCA has: From un total of 64,623 PDB structure files,
30,784 have 'MOLECULAR REPLACEMENT' as Method for Structure Determination.
However, you must remember that we have a large number of false negatives
Several reasons:
- Only 47,557 structure files from the total of 64,623 report which
method was used for structure determination. For example, 1CRY whose
title reports using the MR method does not include the info in the
proper REMARK.
- Users are allowed to write almost anything as the METHOD USED TO
DETERMINE
THE STRUCTURE, making it difficult an accurate report. OCA found PDB
italian
speaking structure files reporting 'MOLECULARE REPLACEMENT' ...
This and other problems are being reported to RCSB."
and one from U of Virginia
"Based on information from Remark200 lines, 31761 structures were solved
using MR, what comprise for 56.8% of all X-RAY structures (55843).
Considering structures which were determined using 'primitive MR' the
number grows to 34949 (62.6%). There are also some structures determined
using combination of MR with SAD, MAD, SIR and MIR. If we would add them,
the number will increase to 35258 (63.1%)."
Thanks again
Nick
