SMILESes for CDK Depict

COP(=O)([O-])OCC P not seen as chiral
[C@](COP(=O)([O-])OCC[N+](C)(C)C)([H])(O)COC=O P seen as chiral
[C@](COP(=O)([O-])OCC[N+](C)(C)C)([H])(OC=O)COC=O P seen as chiral
[C@](COP(=O)([O-])OCC[N+](C)(C)C)([H])(OC(CCCCCCCCCCCCCCCCC)=O)COC(CCCCCCCCCCCCCCCCC)=O P seen as chiral

unittest.java

    @Test
    public void phosphate() throws Exception {
        none(     "COP(=O)([O-])OCC");
        none("[C@](COP(=O)([O-])OCC[N+](C)(C)C)([H])(O-)COC=O");
        none("[C@](COP(=O)([O-])OCC[N+](C)(C)C)([H])(OC=O)COC=O");
        // LMGP01010006 below
        none("[C@](COP(=O)([O-])OCC[N+](C)(C)C)([H])(OC(CCCCCCCCCCCCCCCCC)=O)COC(CCCCCCCCCCCCCCCCC)=O");
    }

Need more info. The P should not be chiral in any of these.

Any none are in CDK depict..

CO[P@](=O)([O-])OCC.[C@](CO[P@](=O)([O-])OCC[N+](C)(C)C)([H])(O-)COC=O.[C@](CO[P@](=O)([O-])OCC[N+](C)(C)C)([H])(OC=O)COC=O.[C@](CO[P@](=O)([O-])OCC[N+](C)(C)C)([H])(OC(CCCCCCCCCCCCCCCCC)=O)COC(CCCCCCCCCCCCCCCCC)=O

All don't show and R/S. Note the wedges will be shown because we put them in, depiction is not standardisation.

Need more info. The P should not be chiral in any of these.

Agreed, but why don't the none() all succeed then?

This is in StereocentersTest.java.

Does this mean that the Stereocenters class just do a first approximation of which atoms are stereo? Or does it return stereogenic? And only when the [O-] of the phosphates get methylated/-ated?

Okay, not True, but Para. Okay, got it. Darn, John, you did so much work. Takes me ages to catch up with everything :(

Okay, reboot. My use of none() was wrong.

So, I rewrote my tests, and I now have these two:

none("P(O)(=O)(OC)OCC");
none("P([OH])(=O)(OC)OCC");
none("P([O-])(=O)(OC)OCC");

The first two pass, but the third fails.

(Or does the InChI code consider the last one a stereocenter??)

I don't know what none() is doing. Where is this?

Nvm found it just was lacking context, you're looking at Stereocenters right?

Yes, https://github.com/cdk/cdk/blob/master/base/test-standard/src/test/java/org/openscience/cdk/stereo/StereocentersTest.java#L811-L815

The thing seems to be this description: https://github.com/cdk/cdk/blob/master/base/standard/src/main/java/org/openscience/cdk/stereo/Stereocenters.java#L80-L83

Here it says that P's with two oxygens (along with some other elements) should not be tetra. The current code (and my none() examples as others in StereocentersTest) indeed finds *-P(=O)(O)-* groups not tetrahedral, but as soon as the hydroxyl group is deprotonated, the current CDK code make is tetrahedral. Is that a bug or a feature (of InChI)?

Ah, you wrote a patch. Thanks!