James,
I wrote a jiffy script for doing this ages ago.
http://bl831.als.lbl.gov/~jamesh/pickup/rmsd
Please feel free to copy/edit/port and use in whatever you like. It's
only a few dozen lines. Reports RMS deviations for all pairs of atoms
with the same "name" (columns 12-27 in the PDB file), no matter how far
apart they are.
I had not noticed Compar before! But it does give me the same result as
my jiffy script if I give it large distance and B cutoffs. Compar might
be faster if that's important.
On a similar subject, reforigin, csymmatch and
phenix.find_alt_orig_sym_mate can do hand flips as well as origin
shifts, but I don't see anything out there except perhaps zanuda that
can hop over alternative indexing choices. For example, if you blindly
solve the same structure from two different data sets and then want to
compare the PDB files, you will get into trouble if your space group is,
say, P31, where there are four different and mutually-incompatible ways
to have indexed the data. Unless you used a reference data set, your
automated data processing program probably chose one of the four
possible indexing conventions at random. Neither Phenix nor CCP4 seem
to have a tool for resolving such cases. Not from PDB files alone.
Yes, you can always go back and re-process your data with a reference,
or you can be more cleverer and use Pointless to re-index the merged
data using a PDB file as a reference, but then you are stuck as to what
real-space operator to apply to the PDB file that has already been
carefully refined against the data that Pointless just re-indexed. What
real-space operator "follows" a given re-indexing operation in
reciprocal space? It's not as trivial as you might think. Yes, you can
always run a Phaser job to avoid thinking about symmetry, but that is a
lot of CPU to burn up just to find a symmetry operator. Zanuda does
this internally, but doesn't seem to have a way of generating
symmetry-allowed alignments without re-refining everything. Or am I
missing something?
-James Holton
MAD Scientist
On 7/3/2017 6:43 AM, R.D. Oeffner wrote:
Hi James
If you have access to a Phenix installation then
phenix.find_alt_orig_sym_mate does what you want in one go. It tries
to position the two MR structures on the same origin and symmetry site
like Csymmatch. Besides computing a score value indicating the quality
of the match it also computes the RMSD between the two structures.
This RMSD is computed with the sequence alignment generated by the SSM
superpose program. So if one MR solution is wrong it would yield a
rather large RMSD value.
Robert
On 03/07/2017 13:16, James Foadi wrote:
Hi Eleanor.
You do the alignment if you are looking to measure the similarity of
two molecules after refinement or the similarity of two molecules in
general. When you do this, like you say, you have to carry out some
sort of alignment, i.e. you move one of the two structures so to have
the highest degree of overlapping with the other , without any
constraint. This is what most of the programs do (and after having
done that they compute the RMSD). But I'm looking for a program to
measure the RMSD of two fixed (un-moved) structures; the structures
have already been moved (with the symmetry and unit cell origin
constrain) by CSYMMATCH. Does it make sense to you?
James
Dr James Foadi PhD Diamond Light Source Ltd. Diamond House Harwell
Science and Innovation Campus Didcot Oxfordshire OX11 0DE United
Kingdom office email: [email protected] alternative email:
[email protected] personal web page: http://www.jfoadi.me.uk
On Monday, 3 July 2017, 11:59, Eleanor Dodson
<[email protected]> wrote:
Dont understand your Q James..
You have to make sort of alignment to get the RMSD surely?
And csymmatch will just apply the symmetry and origin shifts needed
for any comparisonn?
E
On 3 July 2017 at 11:32, Stéphane Duquerroy
<[email protected] <mailto:[email protected]>>
wrote:
Hi James
LSQMAN can calculate the current RMSD between the 2 models
(RMsd_calc mol1 range1 mol2 range2 [Ltarget])
Be careful it renames the chain names
Stephane
------------------------------ ---------------------
Duquerroy Stéphane
Structural Virology Unit - PASTEUR INSTITUTE
25 rue du Dr Roux, 75015 Paris, France
lab: +33 (0)1 45 68 82 66
fax: +33 (0)1 45 68 89 93
email: [email protected] <mailto:[email protected]>
------------------------------ ----------------------
------------------------------------------------------------------------
*De: *"James Foadi" <000009daa8ec3774-dmarc-
[email protected]
<mailto:[email protected]>>
*À: *
*Envoyé: *Lundi 3 Juillet 2017 11:47:00
*Objet: *[ccp4bb] RMSD between unaligned structures
Dear ccp4 tribe,
this might have been asked before, but I haven't paid enough
attention.
I'd like to measure the RMSD between two models after molecular
replacement. I can force the two models to overlap as much as
possible within the symmetry and origin-shift constraints (using
CSYMMATCH). But I don't want the program that compute RMSD to
align the two structures. Can you suggest what I should use? And,
perhaps, what keywords I should adopt?
Many thanks, in advance.
James
Dr James Foadi PhD Diamond Light Source Ltd. Diamond House
Harwell Science and Innovation Campus Didcot Oxfordshire OX11 0DE
United Kingdom office email: [email protected]
<mailto:[email protected]> alternative email:
[email protected] <mailto:[email protected]> personal
web page: http://www.jfoadi.me.uk <http://www.jfoadi.me.uk/>
--
Robert Oeffner, Ph.D.
Research Associate, The Read Group
Department of Haematology,
Cambridge Institute for Medical Research
University of Cambridge
Cambridge Biomedical Campus
Wellcome Trust/MRC Building
Hills Road
Cambridge CB2 0XY
www.cimr.cam.ac.uk/investigators/read/index.html
tel: +44(0)1223 763234
