Align Molecule to template and coordGen not playing nice?

[kulmes]

I'm trying to align different sets of molecules with a common core with the core structure always drawn in the same orientation. I'm doing this by using a template for the core specified as a mol file aligning a rdkit mol generated from a SMARTS string with attachment points specified as *, and then aligning all the target molecules to that rdkit mol (See the code below).

The issue I seem to be running into is that when I turn on coordgen I get roughly the same layout of the core, but atoms around the core have switched positions. I'd prefer to use coordgen as it does do the layout better for the attached groups, and also standardizes bond lengths (my mol file template seems to have shorter bonds than are the default in RDKit) I can fix the bond lengths, but as my molecules get bogger, coordGen does a better job of laying out the attached groups

In the example below, the template specifies the double bond to the sulfur as up, the Florine down, the ether to the right, and any unspecified group to the left. Without Coordgen turned on it correctly lays out SMILES: CC(C)OP(=S)(C)F, with coordgen turned on it still orients the atoms around the P correctly (e.g. it's not tilted as if I just generated the mol directly from the SMILES), but it switches the order, or layout of the atoms around the P.

Example Code:
Template_test.mol:

MJ231400                      

  6  5  0  0  0  0  0  0  0  0999 V2000
   -0.1785    2.4776    0.0000 P   0  0  0  0  0  0  0  0  0  0  0  0
   -0.1785    1.6526    0.0000 F   0  0  0  0  0  0  0  0  0  0  0  0
   -0.1785    3.3026    0.0000 S   0  0  0  0  0  0  0  0  0  0  0  0
    0.6464    2.4776    0.0000 O   0  0  0  0  0  0  0  0  0  0  0  0
   -1.0035    2.4776    0.0000 *   0  0  0  0  0  0  0  0  0  0  0  0
    1.1984    3.0907    0.0000 *   0  0  0  0  0  0  0  0  0  0  0  0
  1  2  1  0  0  0  0
  1  3  2  0  0  0  0
  1  4  1  0  0  0  0
  4  6  1  0  0  0  0
  1  5  1  0  0  0  0
M  END

Python Code:

Chem.rdDepictor.SetPreferCoordGen(True) #Template layout with coordGen doesn't work ... the P arrangement gets messed up
orient = Chem.MolFromMolFile('Template_test.mol', sanitize=False) 
template = Chem.MolFromSmarts('FP(=S)(O[*])[*]') #Note this should match the template with [*] connections as R groups
AllChem.GenerateDepictionMatching2DStructure(template, orient)
m=Chem.inchi.MolFromInchi('InChI=1/C7H16FOPS/c1-6(7(2,3)4)9-10(5,8)11/h6H,1-5H3')
AllChem.GenerateDepictionMatching2DStructure(m,template)
m 

This is what I get with coordgen turned off, correct core, but additional structure isn't the best (although not bad for this simple example ... much worse for a lot of the more complicated molecules)

And this is what I get when turning coordgen on, core gets messed up, but side chain looks better:

For reference this is what I get going directly from an InChI:

So coordgen seems to be partially obeying the template in terms of aligning the atoms attached to P directly up and down and side by side.