New molecule drawer doesn't work properly.

[faribas]

My RMG-Py job crashed with this error:

File "/home/fariba/Code/RMG-Py/rmg.py", line 136, in <module>
    cProfile.runctx(command, global_vars, local_vars, stats_file)
  File "/home/fariba/lib/python2.7/cProfile.py", line 49, in runctx
    prof = prof.runctx(statement, globals, locals)
  File "/home/fariba/lib/python2.7/cProfile.py", line 140, in runctx
    exec cmd in globals, locals
  File "<string>", line 1, in <module>
  File "/home/fariba/Code/RMG-Py/rmgpy/rmg/main.py", line 327, in execute
    self.saveEverything()
  File "/home/fariba/Code/RMG-Py/rmgpy/rmg/main.py", line 471, in saveEverything
    self.saveOutputHTML()
  File "/home/fariba/Code/RMG-Py/rmgpy/rmg/main.py", line 613, in saveOutputHTML
    saveOutputHTML(os.path.join(self.outputDirectory, 'output.html'), self.reactionModel)
  File "/home/fariba/Code/RMG-Py/rmgpy/rmg/output.py", line 95, in saveOutputHTML
    MoleculeDrawer().draw(spec.molecule[0], 'png', fstr)
  File "/home/fariba/Code/RMG-Py/rmgpy/molecule/draw.py", line 207, in draw
    self.__generateCoordinates()
  File "/home/fariba/Code/RMG-Py/rmgpy/molecule/draw.py", line 373, in __generateCoordinates
    self.__generateNeighborCoordinates(backbone)
  File "/home/fariba/Code/RMG-Py/rmgpy/molecule/draw.py", line 697, in __generateNeighborCoordinates
    self.__generateFunctionalGroupCoordinates(atom0, atom1)
  File "/home/fariba/Code/RMG-Py/rmgpy/molecule/draw.py", line 819, in __generateFunctionalGroupCoordinates
    self.__generateFunctionalGroupCoordinates(atom1, atom)
  File "/home/fariba/Code/RMG-Py/rmgpy/molecule/draw.py", line 819, in __generateFunctionalGroupCoordinates
    self.__generateFunctionalGroupCoordinates(atom1, atom)
  File "/home/fariba/Code/RMG-Py/rmgpy/molecule/draw.py", line 760, in __generateFunctionalGroupCoordinates
    center += coordinates_cycle[atoms.index(atom),:]
TypeError: 'NoneType' object has no attribute '__getitem__'

Apparently, the new molecule drawer can't draw levoglucosan " C12C(C(C(C(O2)OC1)O)O)O ", drawing this molecule will fail with the same error message:

xception Value: 'NoneType' object has no attribute '__getitem__'
Traceback:
File "/export/thishost/comocheng/virtualenvs/rmgweb/lib/python2.7/site-packages/django/core/handlers/base.py" in get_response
  111.                         response = callback(request, *callback_args, **callback_kwargs)
File "/export/thishost/comocheng/RMG-website/rmgweb/main/views.py" in drawMolecule
  208.     surface, cr, rect = MoleculeDrawer().draw(molecule, format='png')
File "/export/thishost/comocheng/RMG-website/../RMG-Py/rmgpy/molecule/draw.py" in draw
  207.             self.__generateCoordinates()
File "/export/thishost/comocheng/RMG-website/../RMG-Py/rmgpy/molecule/draw.py" in __generateCoordinates
  373.         self.__generateNeighborCoordinates(backbone)
File "/export/thishost/comocheng/RMG-website/../RMG-Py/rmgpy/molecule/draw.py" in __generateNeighborCoordinates
  697.                         self.__generateFunctionalGroupCoordinates(atom0, atom1)
File "/export/thishost/comocheng/RMG-website/../RMG-Py/rmgpy/molecule/draw.py" in __generateFunctionalGroupCoordinates
  819.                     self.__generateFunctionalGroupCoordinates(atom1, atom)
File "/export/thishost/comocheng/RMG-website/../RMG-Py/rmgpy/molecule/draw.py" in __generateFunctionalGroupCoordinates
  819.                     self.__generateFunctionalGroupCoordinates(atom1, atom)
File "/export/thishost/comocheng/RMG-website/../RMG-Py/rmgpy/molecule/draw.py" in __generateFunctionalGroupCoordinates
  760.                 center += coordinates_cycle[atoms.index(atom),:]

[faribas]

Thanks Josh for your help but my job crashed again with this error message:

File "/home/fariba/Code/RMG-Py/rmg.py", line 136, in <module>
    cProfile.runctx(command, global_vars, local_vars, stats_file)
  File "/home/fariba/lib/python2.7/cProfile.py", line 49, in runctx
    prof = prof.runctx(statement, globals, locals)
  File "/home/fariba/lib/python2.7/cProfile.py", line 140, in runctx
    exec cmd in globals, locals
  File "<string>", line 1, in <module>
  File "/home/fariba/Code/RMG-Py/rmgpy/rmg/main.py", line 393, in execute
    self.saveEverything()
  File "/home/fariba/Code/RMG-Py/rmgpy/rmg/main.py", line 471, in saveEverything
    self.saveOutputHTML()
  File "/home/fariba/Code/RMG-Py/rmgpy/rmg/main.py", line 613, in saveOutputHTML
    saveOutputHTML(os.path.join(self.outputDirectory, 'output.html'), self.reactionModel)
  File "/home/fariba/Code/RMG-Py/rmgpy/rmg/output.py", line 95, in saveOutputHTML
    MoleculeDrawer().draw(spec.molecule[0], 'png', fstr)
  File "/home/fariba/Code/RMG-Py/rmgpy/molecule/draw.py", line 203, in draw
    self.__findRingGroups()
  File "/home/fariba/Code/RMG-Py/rmgpy/molecule/draw.py", line 302, in __findRingGroups
    self.ringSystem.append(cycle)
AttributeError: MoleculeDrawer instance has no attribute 'ringSystem'

[rwest]

I think this is now fixed (by one or two of sean's commits)

[stroiano]

Yes, my commit (which I reverted last night to try to match Josh's since he has a way better understanding of the situation; copy here: 8a44adc) does prevent this exception from being raised, but I don't believe it's a proper fix of the underlying issue.

From __findRingGroups:

# Find all of the cycles in the molecule
self.cycles = self.molecule.getSmallestSetOfSmallestRings()
self.ringSystems = []

# If the molecule contains cycles, find them and group them
if len(self.cycles) > 0:
    # Split the list of cycles into groups
    # Each atom in the molecule should belong to exactly zero or one such groups
    for cycle in self.cycles:
        found = False
        for ringSystem in self.ringSystems:
            for ring in ringSystem:
                if any([atom in ring for atom in cycle]) and not found:
                    self.ringSystem.append(cycle)
                    found = True
        if not found:
            self.ringSystems.append([cycle])

Changing the line self.ringSystem.append(cycle) to self.ringSystems.append([cycle]) does indeed prevent this error (since there is no self.ringSystem), but logically it just doesn't seem right; the result is that every element in self.ringSystems ends up being a list of just one cycle.

It seems to me that the most sensible line here (which I believe is what Josh meant to write in the first place) is just ringSystem.append(cycle) (added in aab0f42). However, this then leads to:

ERROR:root:Error while drawing molecule C1CCC2=C(C1)CCCC2: <Atom 'C'> is not in list
Traceback (most recent call last):
  File "rmgpy/molecule/draw.py", line 207, in draw
    self.__generateCoordinates()
  File "rmgpy/molecule/draw.py", line 334, in __generateCoordinates
    self.__generateRingSystemCoordinates(backbone)
  File "rmgpy/molecule/draw.py", line 500, in __generateRingSystemCoordinates
    center1 += coordinates[atoms.index(atom),:]
ValueError: <Atom 'C'> is not in list

Clearly there's something else wrong here, but I don't know the code well enough yet to make any more guesses.

Also of note though, the other line that I changed in my initial commit was setting __generateRingSystemCoordinates to return coordinates. This is probably not the intended behavior of the function though, as it appears to update self.coordinates appropriately.

Thoughts, @jwallen?

[stroiano]

Alright, things make a bit more sense to me now. This last error stems from atoms becoming blank at some point for any ringSystem with multiple cycles. Through changing three lines, I was able to generate a decent (though slightly distorted) result:

Our web server appears to be randomly flipping one of the rings for some reason, so the above image may look strange here. This is the consistent result that I get locally though:

I don't want to commit this just yet since I've already been making a mess lately as it is, but here's the diff:

diff --git a/rmgpy/molecule/draw.py b/rmgpy/molecule/draw.py
index 0eac468..e9a5bc5 100644
--- a/rmgpy/molecule/draw.py
+++ b/rmgpy/molecule/draw.py
@@ -497,7 +497,7 @@ class MoleculeDrawer:
                 if found:
                     center1 = numpy.zeros(2, numpy.float64)
                     for atom in cycle1:
-                        center1 += coordinates[atoms.index(atom),:]
+                        center1 += coordinates[cycle1.index(atom),:]
                     center1 /= len(cycle1)
                     center0 += center1
                     count += 1
@@ -542,15 +542,16 @@ class MoleculeDrawer:
                 radius = numpy.linalg.norm(center - coordinates[index0,:])

             startAngle = 0.0; endAngle = 0.0
             if len(commonAtoms) == 1:
                 # We will use the full 360 degrees to place the other atoms in the cycle
                 startAngle = math.atan2(-vector[1], vector[0])
                 endAngle = startAngle + 2 * math.pi
             elif len(commonAtoms) >= 2:
                 # Divide other atoms in cycle equally among unused angle
-                vector = coordinates[atoms.index(commonAtoms[-1]),:] - center
+                vector = coordinates[cycle.index(commonAtoms[-1]),:] - center
                 startAngle = math.atan2(vector[1], vector[0])
-                vector = coordinates[atoms.index(commonAtoms[0]),:] - center
+                vector = coordinates[cycle.index(commonAtoms[0]),:] - center
                 endAngle = math.atan2(vector[1], vector[0])

Looking at some of the polycyclics near the bottom of this page shows the result is still pretty bizarre, but at least they draw without error.