[MMTK] pb with nma overlap

[Gaelle Debret]

Hi,

I use NMA overlap to find which modes better describe transition between 
two conformations of a protein. Results I get are strange and different 
from those obtained with webnm@ :

mode     webnma@ overlap       calculated overlap
-------------------------------------------------
1.0   1.4820297314289798e-24   0.00131483002737
2.0   4.3166530471988337e-25   0.0210220093182
3.0   3.1823067040761719e-24   0.00342677193337
4.0   3.0221306094285648e-24   0.0265478032375
5.0   3.4922349400066977e-25   0.0144892041836
6.0   8.5556456104409209e-24   0.00511426131374
7.0   0.10873190750794366      0.245845528549
8.0   0.01466946405600426      0.307636882726
9.0   0.22898898175464505      0.472713296235
10.0  0.13951031951966972      0.281422349579
11.0  0.02686343007541454      0.193467668361
12.0  0.096183683252803576     0.34396879925
13.0  0.0092898852367248111    0.0577602897826
14.0  0.015282384663469546     0.0816245359216
15.0  0.00032720689444287726   0.0433155969729
16.0  0.00049456474482750172   0.0672834796817
17.0  0.0049279130271499697    0.0240859634716
18.0  0.0027433295072599147    0.00770688638589
19.0  0.019275616311892167     0.187232039636
20.0  0.022315112816588413     0.143156894938
21.0  5.2201179313156321e-05   0.0259666337533
22.0  0.0062050927176515525    0.0336032944872
23.0  0.00064686469994443296   0.0829656302871
24.0  0.0035814008682881642    0.0722403029075
25.0  0.0034387668083050208    0.0634913973494
26.0  0.014059475637448568     0.0229432305726
...

Here is the script I use :
(NB: structures have already been fitted with profit)

from MMTK import *
from MMTK.Proteins import Protein
from MMTK.ForceFields import DeformationForceField
from MMTK.NormalModes import NormalModes
from MMTK.PDB import PDBConfiguration
from Gnuplot import plot
from Numeric import sqrt

universe = InfiniteUniverse(DeformationForceField())
molecule = Protein('1cllmini.pdb', model='calpha')
universe.addObject(molecule)

modes = NormalModes(universe)

straight_conf = copy(universe.configuration())
pdb = PDBConfiguration('1cdlmini.pdb')
pdb.applyTo(molecule)
kink_conf = copy(universe.configuration())

vec = kink_conf - straight_conf

overlap = []
for i in range(len(modes)):
     m = modes[i]
     o = m.dotProduct(vec) / \
	sqrt(m.dotProduct(m)) / \
	sqrt(vec.dotProduct(vec))
     overlap.append(abs(o))

Is there something wrong or missing ?
Thanks in advance

Gaelle

-- 
_________________________________________________

  Gaelle DEBRET            debret at ebgm.jussieu.fr
  Equipe Bioinformatique           INSERM - U 726
  Genomique et Moleculaire     Case Courrier 7113
  2, place Jussieu,	     tel: 01 44 27 63 50
  75251 Paris Cedex 05        fax: 01 43 26 38 30
_________________________________________________