relax SVN archive

Clean ups and improvements for the pipe_control.rdc.check_rdcs() function.

Pseudo-atoms are now handled much better and correctly in all cases. And many RelaxErrors have been
converted to RelaxWarnings followed by a 'return False' statement.

Added some RDC data checks to the N_state_model.test_pyrotartaric_anhydride_rdcs system test.

This is to demonstrate a problem with the data assembly function pipe_control.rdc.return_rdc_data().

Create two new warnings RelaxNucleusWarning and RelaxSpinTypeWarning to match the equivalent errors.

Movement of N-state model specific code to the analysis neutral pipe_control package.

Many of the functions of the specific_analyses.n_state_model.data module relating to alignment
tensors, RDC data and PCS data have been shifted in to the pipe_control package modules
align_tensor, rdc, and pcs respectively. This allows these functions to be made more general and
allow the code to be shared with the frame order analysis or any future analysis using such data,
and hence remove some code duplication.

Created the N_state_model.test_pyrotartaric_anhydride_mix system test.

This is used to demonstrate a bug in the N-state analysis using mixed RDC and long range absolute
J+D data.

Merged revisions 21391,21407-21434 via svnmerge from
svn+ssh://bugman@.../svn/relax/trunk

........
r21391 | bugman | 2013-11-06 10:21:38 +0100 (Wed, 06 Nov 2013) | 7 lines

Improvements to the pymol.view and molmol.view user functions for finding the PDB files.

Now the possibility that this is being run from a results subdirectory is taken into consideration.
If the file cannot be found, the os.pardir parent directory is added to the start of the relative
path and the file checked for.
........
r21407 | bugman | 2013-11-12 16:47:16 +0100 (Tue, 12 Nov 2013) | 5 lines

The rdc.read user function will now skip all lines of the RDC file starting with '#'.

To include molecule identifiers at the start of the line will now require quotation marks.
........
r21408 | bugman | 2013-11-12 19:35:23 +0100 (Tue, 12 Nov 2013) | 3 lines

Shifted the RDC and PCS assembly methods from the main class to the data module for the N-state analysis.
........
r21409 | bugman | 2013-11-12 19:45:04 +0100 (Tue, 12 Nov 2013) | 3 lines

Created the pipe_control.mol_res_spin.is_pseudoatom() function to simplify pseudo-atom handling.
........
r21410 | bugman | 2013-11-12 19:54:20 +0100 (Tue, 12 Nov 2013) | 5 lines

Created the pipe_control.mol_res_spin.pseudoatom_loop() function.

This is used to loop over the spin containers corresponding to a given pseudo-atom.
........
r21411 | bugman | 2013-11-13 09:57:41 +0100 (Wed, 13 Nov 2013) | 5 lines

Added a PDB file and RDC values (and absolute J+D and J) for propylene carbonate.

This will be used for testing of pseudo-atoms in the N-state model analysis.
........
r21412 | bugman | 2013-11-13 10:04:28 +0100 (Wed, 13 Nov 2013) | 3 lines

Renamed the propylene carbonate files to the correct name of pyrotartaric anhydride.
........
r21413 | bugman | 2013-11-13 10:47:37 +0100 (Wed, 13 Nov 2013) | 8 lines

Created two new system tests based on the new pyrotarctic anhydride long range (1J, 2J & 3J) RDC data.

The first (N_state_model.test_pyrotartaric_anhydride_rdcs) optimises an alignment tensor using long
range signed RDC data. The second (N_state_model.test_pyrotartaric_anhydride_absT) optimises an
alignment tensor using long range absolute T (J+D) data. Both test long range data together with
methyl group pseudo-atom data.
........
r21414 | bugman | 2013-11-13 11:06:59 +0100 (Wed, 13 Nov 2013) | 5 lines

Added all of the pyrotartaric anhydride RDC generation scripts and files.

This is simply for reference and reproducibility.
........
r21415 | bugman | 2013-11-13 11:09:09 +0100 (Wed, 13 Nov 2013) | 5 lines

Modifications for the pyrotartaric anhydride system test script.

The grid search now is much quicker, and the RDC correlation plots are now sent to DEVNULL.
........
r21416 | bugman | 2013-11-13 11:12:12 +0100 (Wed, 13 Nov 2013) | 6 lines

Added the return_id argument to the pipe_control.mol_res_spin.pseudoatom_loop() function.

This will then yield both the spin container and spin ID string. This mimics the spin_loop()
function.
........
r21417 | bugman | 2013-11-13 14:05:03 +0100 (Wed, 13 Nov 2013) | 12 lines

Added proper pseudo-atom support for the RDCs in the N-state model analysis.

This involves a number of changes. The pseudo-atom specific functions ave_rdc_tensor_pseudoatom()
and ave_rdc_tensor_pseudoatom_dDij_dAmn() have been added to the lib.alignment.rdc module. These
simply average the values from the equivalent non-pseudo-atom functions. The return_rdc_data()
function in the specific_analyses.n_state_model.data module has been modified to assemble the RDC
constants and unit vectors for all members of the pseudo-atom and add these to the returned
structures, as well as a new list of flags specifying if the interatom pair contains pseudo-atoms.
The N-state model target function and gradient have been updated to send the pseudo-atom data to the
new lib.alignment.rdc module functions.
........
r21418 | bugman | 2013-11-13 14:18:01 +0100 (Wed, 13 Nov 2013) | 6 lines

J couplings for the N-state analysis are now properly handled for pseudo-atoms.

The measured J couplings for the members of the pseudo-atom should not be used, but rather that of
the pseudo-atom spin itself (as the former does not exist).
........
r21419 | bugman | 2013-11-13 14:26:23 +0100 (Wed, 13 Nov 2013) | 6 lines

Fix for the pyrotartaric_anhydride.py N-state model system test script.

This is used in two system tests. The user function call for reading the inter-spin distances was
instead reading the J couplings as the distance.
........
r21420 | bugman | 2013-11-13 14:58:54 +0100 (Wed, 13 Nov 2013) | 3 lines

The interatomic loop now skips the element if either of the spin systems are deselected.
........
r21421 | bugman | 2013-11-13 15:02:27 +0100 (Wed, 13 Nov 2013) | 6 lines

Eliminated the old pseudo-atom handling in the N-state model specific return_rdc_data() function.

This was multiplying the RDCs by -3 to handle the tetrahedral geometry of the 1J methyl RDCs.
However this approach is not valid for non-methyl pseudo-atoms or for 2J, 3J, etc. data.
........
r21422 | bugman | 2013-11-13 15:06:50 +0100 (Wed, 13 Nov 2013) | 17 lines

Reverted r21420 as the spin containers and interatomic data containers should be handled separately.

The command used was:
svn merge -r21420:r21419 .

This is important to allow interatomic containers with good RDC data to be selected while one of the
spin pairs deselected (for example if its PCS data is bad).

.....
r21420 | bugman | 2013-11-13 14:58:54 +0100 (Wed, 13 Nov 2013) | 3 lines
Changed paths:
M /trunk/pipe_control/interatomic.py

The interatomic loop now skips the element if either of the spin systems are deselected.
.....
........
r21423 | bugman | 2013-11-13 15:33:18 +0100 (Wed, 13 Nov 2013) | 6 lines

A RelaxError is now raised for the N-state model optimisation with gradients when T = J+D data is used.

The gradients for this data type are not implemented yet, so it is better to prevent the user from
using this.
........
r21424 | bugman | 2013-11-13 15:34:40 +0100 (Wed, 13 Nov 2013) | 5 lines

The N_state_model.test_pyrotartaric_anhydride_absT system test now uses simplex optimisation to pass.

The Newton algorithm cannot be used as the gradients for T = J+D type data have not been implemented.
........
r21425 | bugman | 2013-11-13 16:51:00 +0100 (Wed, 13 Nov 2013) | 5 lines

An RDC error of 0.0 will now deselect the corresponding interatomic data container.

This can be used for simpler pseudo-atom handling.
........
r21426 | bugman | 2013-11-13 16:51:59 +0100 (Wed, 13 Nov 2013) | 3 lines

Updated the menthol long range RDC data file to include pseudo-atom member distances.
........
r21427 | bugman | 2013-11-13 16:59:10 +0100 (Wed, 13 Nov 2013) | 5 lines

Renamed the interatomic_loop() function 'selected' argument to 'skip_desel'.

This is to match the spin_loop() function arguments.
........
r21428 | bugman | 2013-11-13 17:02:06 +0100 (Wed, 13 Nov 2013) | 6 lines

The interatom.unit_vectors user function now calculates the unit vectors for deselected containers.

This is useful for pseudo-atom handling where the interatomic containers to the pseudo-atom members
have already been deselected.
........
r21429 | bugman | 2013-11-13 17:05:05 +0100 (Wed, 13 Nov 2013) | 5 lines

Updated the value checking for the N_state_model.test_absolute_rdc_menthol system test.

The pseudo-atoms are now properly handled so the result is now much better.
........
r21430 | bugman | 2013-11-13 17:35:30 +0100 (Wed, 13 Nov 2013) | 5 lines

The stereochemistry auto-analysis can now accept a file of interatomic distances.

This is for better pseudo-atom support.
........
r21431 | bugman | 2013-11-13 17:46:19 +0100 (Wed, 13 Nov 2013) | 3 lines

The N-state model specific check_rdcs() function now properly handles pseudo-atoms.
........
r21432 | bugman | 2013-11-13 17:52:22 +0100 (Wed, 13 Nov 2013) | 5 lines

The pipe_control.rdc.q_factors() function now properly handles pseudo-atoms.

If pseudo-atoms are present, then 2Da^2(4 + 3R)/5 normalised Q factor is skipped.
........
r21433 | bugman | 2013-11-13 18:12:05 +0100 (Wed, 13 Nov 2013) | 5 lines

Fix for the N_state_model.test_stereochem_analysis system test for the pseudo-atom handling.

An interatomic distance file is now used to explicitly set up all distances.
........
r21434 | bugman | 2013-11-13 18:30:44 +0100 (Wed, 13 Nov 2013) | 6 lines

Fixes for the N-state model target function gradient and Hessian for the data structure changes.

The unit vector data structure is no longer a pure numpy array but a mixed list and array structure
to handle pseudo-atoms.
........

Fixes for the N-state model target function gradient and Hessian for the data structure changes.

The unit vector data structure is no longer a pure numpy array but a mixed list and array structure
to handle pseudo-atoms.

Fix for the N_state_model.test_stereochem_analysis system test for the pseudo-atom handling.

An interatomic distance file is now used to explicitly set up all distances.

The pipe_control.rdc.q_factors() function now properly handles pseudo-atoms.

If pseudo-atoms are present, then 2Da^2(4 + 3R)/5 normalised Q factor is skipped.

The N-state model specific check_rdcs() function now properly handles pseudo-atoms.