Hi Ed,
I started converting the consistency tests code to the new design.
However, it is mostly possible that I make huge mistakes and/or forget
many things.
So could you tell me if what I've done so far (r4583 of the
consistency_tests branch) is fine ?
Cheers,
Sébastien :)
Edward d'Auvergne wrote:
There are a number of significant changes required, so don't feel
obliged to do this. Maybe if I make the changes to the reduced spectral
density mapping code first, you can then see what is needed. But if you
would like to make this code fully functional in the 1.3 line and would
like a challenge, then the run argument is the most important of these.
The 'run' no longer exists, it is now replaced by the concept of the
data pipe or simply pipe. The mode of operation of relax is that there
is a current pipe (stored in 'relax_data_store.current_pipe') and that
that is what the code should access. So for the functions that require
access to the data structures of this data pipe, a common piece of code
I have been using is:
# Alias the current data pipe.
cdp = relax_data_store[relax_data_store.current_pipe]
The structure 'cdp' then contains all the data you need (technically it
is a PipeContainer instance). The only functions which should require a
data pipe other than the current one are things like copying between
data pipes, model selection, and hybridization of data pipes.
Another change is:
# Test the sequence data exists:
if not relax_data_store.res.has_key(run):
raise RelaxNoSequenceError, run
to:
# Test if the sequence data is loaded.
if not exists_mol_res_spin_data():
raise RelaxNoSequenceError
Looping over residues should now not be done. So the code:
for i in xrange(len(relax_data_store.res[self.run])):
# Skip unselected residues.
if not relax_data_store.res[self.run][i].select:
continue
should be replaced by the spin loop:
for spin in spin_loop(spin_id):
# Skip unselected spins.
if not spin.select:
continue
The spin_id (spin identification string) I don't think will be required
by the consistency tests. As you can probably see the spin_loop makes
the code much simpler. The epydoc '@param x:' and '@type x:' parts of
the docstring would be useful as well. Oh, I also just noticed that
there is a tab in one place in specific_fns/consistency_tests.py. This
is in the set_doc() method where something strange has happened. I
think some lines and the start of a function have been lost somewhere.
What else is there? Oh, all functions which previously accepted the
'index' argument should be changed. Instead the SpinContainer instance
(the object returned by the spin loop) should be passed in instead.
Then all lines such as:
if not hasattr(relax_data_store.res[self.run][index], 'csa') or not
hasattr(relax_data_store.res[self.run][index], 'r') or not
hasattr(self.relax.data.res[self.run][index], 'orientation') or not
hasattr(self.relax.data.res[self.run][index], 'tc'):
self.data_init(relax_data_store.res[self.run][index])
should be replace by something such as:
if not hasattr(spin, 'csa') or not hasattr(spin, 'r') or not
hasattr(spin, 'orientation') or not hasattr(spin, 'tc'):
self.data_init(spin)
For the writing of results files, this will be more difficult as a few
extra columns will be necessary. Maybe it's best if I implement the
idea somewhere else first, because I'm not sure exactly what I want here
yet. In the 1.2 line there was a residue number and residue name
column. In the 1.3 line there should be 5 columns: 1st, mol name; 2nd,
res num; 3rd, res name; 4th, spin num; 5th, spin name. These can be
accessed from the spin_loop() function (well iterator actually) by
setting the full_info argument to True. For most protein systems the
molecule and spin columns are likely to be filled by None, so I'm not
sure whether this info should be pre-parsed and columns dropped if there
is no info. For example for a small organic molecule, the residue info
is likely to be useless. For protein NH backbone data, the spin columns
are useless. For single molecule systems the molecule column is
useless. I will drop these columns when writing other data to file, but
in the results file maybe it's ok to keep them. I'm not sure yet.
I think that comprehensively covers everything! If you see anything
else, don't hesitate to ask. Don't forget that unit tests will uncover
almost every last bug in the code. Also for the system tests, if you
have a small data subset and a consistency testing script to analyse
this data then you can include these directly now. The data must be
very small in size if it is to be distributed together with relax.
Regards,
Edward
On Tue, 2008-01-08 at 13:31 -0500, Sebastien Morin wrote:
Hi,
Could you tell me what will be the change so I can try to make it for
the consistency_test code and, if good, port it to the jw_mapping
code ?
Maybe some old posts in the mailing lists talked about this...
Cheers
Séb
Edward d'Auvergne wrote:
An important note here is that the run arguments will be removed from
the 1.3 line. I haven't done this for the J(w) mapping code yet
though. I'm slowly converting the whole code base to remove this
argument, so I should get to it one day soon.
Regards,
Edward
On Jan 7, 2008 10:17 PM, <sebastien.morin.1@xxxxxxxxx> wrote:
Author: semor
Date: Mon Jan 7 22:17:10 2008
New Revision: 4467
URL: http://svn.gna.org/viewcvs/relax?rev=4467&view=rev
Log:
Modified the consistency tests prompt so it is not a simple copy of the jw mapping code anymore.
Modified:
branches/consistency_tests_1.3/prompt/consistency_tests.py
Modified: branches/consistency_tests_1.3/prompt/consistency_tests.py
URL: http://svn.gna.org/viewcvs/relax/branches/consistency_tests_1.3/prompt/consistency_tests.py?rev=4467&r1=4466&r2=4467&view=diff
==============================================================================
--- branches/consistency_tests_1.3/prompt/consistency_tests.py (original)
+++ branches/consistency_tests_1.3/prompt/consistency_tests.py Mon Jan 7 22:17:10 2008
@@ -27,11 +27,12 @@
from relax_errors import RelaxStrError
-class Jw_mapping:
+class Consistency_tests:
def __init__(self, relax):
# Help.
self.__relax_help__ = \
- """Class containing functions specific to reduced spectral density mapping."""
+ """Class containing functions specific to consistency tests for datasets from different
+ fields."""
# Add the generic help string.
self.__relax_help__ = self.__relax_help__ + "\n" + help.relax_class_help
@@ -41,7 +42,7 @@
def set_frq(self, run=None, frq=None):
- """Function for selecting which relaxation data to use in the J(w) mapping.
+ """Function for selecting which relaxation data to use in the consistency tests.
Keyword Arguments
~~~~~~~~~~~~~~~~~
@@ -54,20 +55,20 @@
Description
~~~~~~~~~~~
- This function will select the relaxation data to use in the reduced spectral density mapping
- corresponding to the given frequency.
+ This function will select the relaxation data to use in the consistency tests corresponding
+ to the given frequencies.
Examples
~~~~~~~~
- relax> jw_mapping.set_frq('jw', 600.0 * 1e6)
- relax> jw_mapping.set_frq(run='jw', frq=600.0 * 1e6)
+ relax> consistency_tests.set_frq('ct', 600.0 * 1e6)
+ relax> consistency_tests.set_frq(run='ct', frq=600.0 * 1e6)
"""
# Function intro text.
if self.__relax__.interpreter.intro:
- text = sys.ps3 + "jw_mapping.set_frq("
+ text = sys.ps3 + "consistency_tests.set_frq("
text = text + "run=" + `run`
text = text + ", frq=" + `frq` + ")"
print text
@@ -81,4 +82,4 @@
raise RelaxStrError, ('frq', frq)
# Execute the functional code.
- self.__relax__.specific.jw_mapping.set_frq(run=run, frq=frq)
+ self.__relax__.specific.consistency_tests.set_frq(run=run, frq=frq)
_______________________________________________
relax (http://nmr-relax.com)
This is the relax-commits mailing list
relax-commits@xxxxxxx
To unsubscribe from this list, get a password
reminder, or change your subscription options,
visit the list information page at
https://mail.gna.org/listinfo/relax-commits
--
Sebastien Morin
Etudiant au PhD en biochimie
Laboratoire de resonance magnetique nucleaire
Dr Stephane Gagne
CREFSIP (Universite Laval, Quebec, CANADA)
1-418-656-2131 #4530
--
Sebastien Morin
Etudiant au PhD en biochimie
Laboratoire de resonance magnetique nucleaire
Dr Stephane Gagne
CREFSIP (Universite Laval, Quebec, CANADA)
1-418-656-2131 #4530
|
Re: Shifting the consistency testing code to the new relax design.