Hi Aimee,
To get the scaling properly you need something like this:
A=40.000 #set this to the experimental value of your
supercell's a-vector
for frame in range(frames):
a=celllen[frame,0] #get the a-vector of the current frame
scale=a/A #scale factor (pressure scaling is isotropic in Amber
so your scale factor will be the same in all dimensions
currMoveVector=MoveVector*scale # scale your MoveVector
coords[frame,:,:]=coords[frame,:,:]-MoveVector
Try that and see what happens although if you are monitoring your volume
and it is not deviating by more than 0.3% or so from the experimental,
this should really not affect your B-factors all that much.
Besides that I'm really not sure what the reason could be. Possibly the
algorithm for fitting the unit cells (the least squares problem for
finding the correct origin that I mentioned in previous e-mail) could
have made a difference. Possibly your system is sampling an alternate
ensemble that is producing different B-factors.
Pawel
On 10/29/2013 04:03 AM, Yongxiu Li wrote:
> hello,Pawel
> Thank you very much! I am so sorry for that I read the
> RevSym_netcdf.py again and I think I have done the rescaling. I think
> these lines do the all rescaling "
> for frame in range(frames):
> #get frame box
> box=celllen[frame,:]
> #convert to unit cell box
> box=box/[ix,iy,iz]
> #add angle information as provided by user
> UCbox=hstack((box,SCBox[3:]))
> #calc orthogonalization matrix
> u,invu=CompXfrm(UCbox)
> #matrix product
> MoveVector=dot(invu,FracVector).astype(float32)
> coords[frame,:,:]=coords[frame,:,:]-MoveVector
> # reverse symmetry operate to original asym unit by
> # applying the symmetry translation and rotation
> coords[:,:,:]=dot( (coords[:,:,:]-t),linalg.inv(s) ) "
> I'm not proficient in Python, so whether I understand is right? If
> not, can you tell me how to do in python script? If I am right, I
> still can't understand that compared with figure8 in Case's paper,why
> the bfac_lat_calpha-arf.dat which I gained is so big.
> btw,in the attachment is the lattice atomic root mean squared
> fluctuations of a-carbons by amber ff99SB-tip3p.
> Thank you very much!
> Aimee Li
>
> Atomicrootmeansquared(rms)fluctuationsofa-carbons.
>
> On 10/16/2013 07:26 AM, Pawel wrote:
>> Hi Aimee,
>> On 10/15/2013 09:09 AM, Yongxiu Li wrote:
>>> 1. if the "bfac_lat_calpha" in figure8 are calculated using a
>>> different
>>> approach than the one used in BasicAnalysis, how can I do it?
>> If I understand correctly (I did not work on that paper myself), the
>> alignment of asymmetric units was done by solving a least squares
>> minimization problem to find (for each frame) the best location of the
>> crystallographic origin in space so that after performing the
>> appropriate symmetry operations and translations in reverse on each of
>> the asymmetric units, the resulting RMSD between the asymmetric units
>> would be minimized. Unfortunately I don't have the code for that
>> approach. If it would be very useful to you, let me know and I'll
>> investigate.
>>
>>> So as you said, I need to rescal coordinates.
>>> However, after reading this paper I am confused that wether only
>>> computing the lattice property (for example:rmsd_latt and
>>> bfactor_latt)need rescal coordinates but computing the other
>>> property(like rmsd_ASU ,bfactor_ASU and Distance deviation matrices)
>>> needn't rescale the coordinates?
>> If you are simulating an NPT ensemble, you should rescale the
>> coordinates for all of these analyses. This should be done in
>> RevSym_netcdf.py though I think the version of that script that you have
>> does not do so. Like I said these scripts Analysis directory are all
>> beta versions. However you should be able to do the rescaling easily by
>> adding one or two line around line 95 of the script. If you have
>> problems, let me know and I'll dig up a version that does this.
>>
>> Pawel
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>>
>
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed Oct 30 2013 - 06:30:04 PDT
