Re: BMRB diffusion tensor saveframes. -- November 19, 2009 - 10:06

Hi,

Yes, the PDB file could be rotated.  This is not done in relax yet,
but might in the future if molecular superimposition is implemented.
There is also the possibility that the PDB file is rotated by the
user, or that it is not actually deposited in the PDB (from an MD
simulation, etc.).  Therefore I could get relax (http://nmr-relax.com)
to dump atomic coordinate information, and add this support to bmrblib
(https://gna.org/svn/?group=bmrblib, again a homepage link because
this doesn't yet appear on the first Google search page!).  The
question is, what should I include.  I could add all atomic position
information, or maybe I could add just the atoms investigated.  The
nearby atoms causing dipolar relaxation could also be added, but this
information might already be contained in the other saveframes
(interatomic_dist, bond_vectors, etc.).  What do you think?

Cheers,

Edward



2009/11/18 Eldon Ulrich <elu@xxxxxxxxxxxxx>:
> Hi,
>
> One quick question; Is it possible that the coordinates from the PDB file
> would be modified (two models superimposed, or some rotation carried out) so
> that the coordinates used for the tensors would not be the original PDB
> coordinates? Another possibility is that the original PDB entry might be
> updated or obsoleted. If any of these, are possible (likely), then the
> coordinates for the models should probably be included in the NMR-STAR file,
> with a reference to the original PDB entry and model number, and a
> description (software used, etc.) for carrying out the transformation.
> > Eldon
>
> Edward d'Auvergne wrote:
> > Eldon,
> >
> > Would you be able to help me with a few questions about the Tensor
> > save frame that I am about to create?  Sorry for the length :S  This
> > save frame, from memory, was a general save frame for all types of
> > rank-2 tensors, was it not?  It could therefore hold diffusion
> > tensors, alignment tensors, Saupe order matrices, CS tensors, etc.?
> > This appears to be an expansion of shielding tensor save frame.  Below
> > are a few suggestions for additions and some mock ups that I will code
> > relax (http://nmr-relax.com) and bmrblib
> > (http://gna.org/projects/bmrblib/) to produce.
> >
> >
> > > _Tensor.Reduceable_matrix_1_1_val
> > > _Tensor.Reduceable_matrix_1_2_val
> > > _Tensor.Reduceable_matrix_1_3_val
> > > _Tensor.Reduceable_matrix_2_1_val
> > > _Tensor.Reduceable_matrix_2_2_val
> > > _Tensor.Reduceable_matrix_2_3_val
> > > _Tensor.Reduceable_matrix_3_1_val
> > > _Tensor.Reduceable_matrix_3_2_val
> > > _Tensor.Reduceable_matrix_3_3_val
> >
> > Firstly, I'm guessing this matrix is in the structure frame (the PDB
> > frame).  But how should we specify the PDB frame?  For example I could
> > be analysing CaM using Ad Bax's high quality, yet different PDB
> > structures for each domain separately, i.e. model 2 of 1J7P and model
> > 3 of 1J7O.  Should I specify PDBs and model numbers, or dump all
> > coordinate info of the structures?  This information is needed because
> > the diffusion tensor is attached to a structure and the structures are
> > in random orientations within the PDB frame.
> >
> > One problem I see is how do we specify the tensor type?  Maybe a tag
> > such as _Tensor_list.Tensor_type with fixed enumerations such as
> > 'diffusion', 'CS', 'shielding', and 'alignment'?  These tensors are
> > often described quite differently - though the
> > _Tensor.Reduceable_matrix_* form applies to all.  For example the
> > total information content of a spheroidal diffusion tensor from relax
> > in XML form looks like:
> >
> >        <diff_tensor desc="Diffusion tensor" type="spheroid">
> >            <Da ieee_754_byte_array="[213, 118, 248, 246, 40, 53, 91, 65]">
> >                7132323.85892
> >            </Da>
> >            <fixed>
> >                False
> >            </fixed>
> >            <phi ieee_754_byte_array="[224, 227, 53, 166, 165, 212, 3,
> > 64]">
> >                2.47883157589
> >            </phi>
> >            <spheroid_type>
> >                'prolate'
> >            </spheroid_type>
> >            <theta ieee_754_byte_array="[147, 6, 168, 205, 63, 251, 238,
> > 63]">
> >                0.968170072243
> >            </theta>
> >            <tm ieee_754_byte_array="[78, 9, 5, 238, 170, 132, 67, 62]">
> >                9.08888558914e-09
> >            </tm>
> >        </diff_tensor>
> >
> > Note that in this axially symmetric case there are, in fact, infinite
> > matrix forms.  Therefore just an axis and tm and Da are used to
> > specify this.  But multiple parameters can be used to specify a
> > diffusion tensor, and different definitions/equations of properties
> > such as rhombicity are used all over the place.  Therefore the _Tensor
> > save frame will require a little refinement, or that this is used as a
> > template for specific tensor save frames.
> >
> > Below are some mock ups of diffusion tensors using the _Tensor save
> > frame, with a few additional tags.  I think I can use a separate save
> > frame for each diffusion tensor, so maybe 2 saveframes for a 2 domain
> > system?  It could probably all go into one frame, but might be nicer
> > to group these into individual frames, especially if tensors overlap!
> > I've added the _Tensor_list.*_val_type tags to handle different
> > definitions of anisotropy and rhombicity as these can be defined in
> > many different ways!  This is the minimal information I need to
> > deposit and then re-read the data.
> >
> >
> > # Spherical diffusion mock up (unused tags removed).
> >
> > save_diff_tensor
> >   _Tensor_list.Sf_category tensor
> >   _Tensor_list.Sf_framecode spherical_tensor_1
> >   _Tensor_list.ID 1
> >   _Tensor_list.Tensor_type diffusion        # New.
> >   _Tensor_list.Matrix_val_units s-1        # New (this is the units
> > of the matrix form, and could be ppm for CS, etc.).
> >   _Tensor_list.Isotropic_val_type "Diso = 1/(6.tm)"        # New.
> >   _Tensor_list.Details "spherical diffusion"
> >
> >   loop_
> >      _Tensor.ID
> >      _Tensor.Entity_ID
> >      _Tensor.Comp_index_ID
> >      _Tensor.Comp_ID
> >      _Tensor.Atom_ID
> >      _Tensor.Atom_type
> >      _Tensor.Atom_isotope_number
> >      _Tensor.Isotropic_val
> >      _Tensor.Anisotropic_val
> >      _Tensor.Rhombic_val
> >      _Tensor.Tensor_list_ID
> >
> > 1 1 9 Gln N N 15 12985705.377075171 0.0 0.0 1
> > 2 1 10 Gln N N 15 12985705.377075171 0.0 0.0 1
> >
> >   stop_
> >
> > save_
> >
> >
> >
> > # Spheroidal diffusion mock up (unused tags removed).
> > # Note that _Tensor_list.Spheroid_type can be determined by the sign
> > of Da (_Tensor.Anisotropic_val) when using Da = Dpar - Dper.
> > # Also note that the polar (theta) and azimuthal angle (phi) are in
> > this example, as in most cases, defined as 0 <= theta <= pi; 0 <= phi
> > <= 2pi.
> >
> > save_diff_tensor
> >   _Tensor_list.Sf_category tensor
> >   _Tensor_list.Sf_framecode spheroidal_tensor_1
> >   _Tensor_list.ID 1
> >   _Tensor_list.Tensor_type diffusion        # New.
> >   _Tensor_list.Matrix_val_units s-1        # New (this is the units
> > of the matrix form, and could be ppm for CS, etc.).
> >   _Tensor_list.Spheroid_type prolate        # New.
> >   _Tensor_list.Angle_units rad        # New (for polar and azimuthal
> > angles, as well as the 3 Euler angles).
> >   _Tensor_list.Isotropic_val_formula "Diso = 1/(6.tm)"        # New.
> >   _Tensor_list.Anisotropic_val_formula "Da = Dpar - Dper"       # New.
> >   _Tensor_list.Details "spheroidal diffusion"
> >
> >   loop_
> >      _Tensor.ID
> >      _Tensor.Entity_ID
> >      _Tensor.Comp_index_ID
> >      _Tensor.Comp_ID
> >      _Tensor.Atom_ID
> >      _Tensor.Atom_type
> >      _Tensor.Atom_isotope_number
> >      _Tensor.Axial_sym_axis_polar_angle        # New.
> >      _Tensor.Axial_sym_axis_azimuthal_angle        # New.
> >      _Tensor.Isotropic_val
> >      _Tensor.Anisotropic_val
> >      _Tensor.Rhombic_val
> >      _Tensor.Tensor_list_ID
> >
> > 1 1 9 Gln N N 15 0.96817007224326956 2.4788315758892034
> > 18337414.970415145 7132323.8589150626 0.0 1
> > 2 1 10 Gln N N 15 0.96817007224326956 2.4788315758892034
> > 18337414.970415145 7132323.8589150626 0.0 1
> >
> >   stop_
> >
> > save_
> >
> >
> >
> > # Ellipsoidal diffusion mock up (unused tags removed).
> >
> > save_diff_tensor
> >   _Tensor_list.Sf_category tensor
> >   _Tensor_list.Sf_framecode ellipsoidal_tensor_1
> >   _Tensor_list.ID 1
> >   _Tensor_list.Tensor_type diffusion        # New.
> >   _Tensor_list.Matrix_val_units s-1        # New (this is the units
> > of the matrix form, and could be ppm for CS, etc.).
> >   _Tensor_list.Angle_units rad        # New (for polar and azimuthal
> > angles, as well as the 3 Euler angles).
> >   _Tensor_list.Euler_angle_type zyz        # New.
> >   _Tensor_list.Isotropic_val_formula "Diso = 1/(6.tm)"        # New.
> >   _Tensor_list.Anisotropic_val_formula "Da = Dpar - Dper"       # New.
> >   _Tensor_list.Rhombic_val_formula "Dr = (Dy - Dx)/2Da"       # New.
> >   _Tensor_list.Details "ellipsoidal diffusion"
> >
> >   loop_
> >      _Tensor.ID
> >      _Tensor.Entity_ID
> >      _Tensor.Comp_index_ID
> >      _Tensor.Comp_ID
> >      _Tensor.Atom_ID
> >      _Tensor.Atom_type
> >      _Tensor.Atom_isotope_number
> >      _Tensor.Euler_angle_alpha
> >      _Tensor.Euler_angle_beta
> >      _Tensor.Euler_angle_gamma
> >      _Tensor.Isotropic_val
> >      _Tensor.Anisotropic_val
> >      _Tensor.Rhombic_val
> >      _Tensor.Tensor_list_ID
> >
> > 1 1 9 Gln N N 15 1.22079138007 2.61076471643 0.632663091701
> > 39314097.38005387 235884584.281 9.58992200396e-17 1
> > 2 1 10 Gln N N 15  1.22079138007 2.61076471643 0.632663091701
> > 39314097.38005387 235884584.281 9.58992200396e-17 1
> >
> >   stop_
> >
> > save_
> >
> >
> >
> > Ok, that should be hopefully it for this save frame.  When I can work
> > on it, I'll get relax and bmrblib to produce these mock ups.
> >
> > Cheers,
> >
> > Edward