I found in some publication, that the authors calculated the log P values for the platinum(II) complexes, using Marvin Sketch. Now, I am trying to do the same thing, but is it now possible. Is it some kind of problems with my software or the authors lied??
Could you identify the reference at issue?
Now, I am trying to do the same thing, but is it now possible.
I was able to reproduce the calculated logP values which are given in the "table 1" of the referenced paper.
For example the calculated logP of oxaliplatin is 1.73 as it is shown in the paper.
The right input structure of the oxaliplatin in MarvinSketch and the calculated logP value is given on the attached figure.
but don't you forget about coordination bonds between platinum ion and nitrogens??
Please, look at this small window. Due to the lack of coordination bonds, there are NH groups in DACH instead of NH2. So the structure is incorrect.
No, it is done intentionally because of the coordination bond which is not accepted by the present version of the calculators.
Calculation will break in case a coordination bond drawn in the structure.
This is why a simple single bond was drawn instead of the "arrow bond " on the attached MarvinSketch's figure between the lone pair donors and the metal center.
This is the "workaround" for handling of the coordination bond. Later this shortcoming will be fixed and the "H" atom count will be correctly adjusted too.
Now I understand everything. Thanks for your explanation. It is good to know that I can use log P prediction in my research.
My question related to this thread:
I would like to predict the log P / log D for a coordination complex with keto groups as donors. When I draw a regular bond from the ketone oxygen to the metal Marvin assigns a negative charge to O and does not allow to predict the log P because it considers it an "Inconsistent molecular structure".
In the example above this problem did not occur because the amino donor had a proton that could be omitted to draw the bond without a negative bond to appear. How should I proceed? Using coordination bonds does not work either, as discussed above.
Do I need to use partial charges Pt(2+) and O(-) for the ketone, or should I even draw the ketone with only a single bond (C--O--M) instead of (C==O--M)? This way the simulation works but I wonder how reliable this is in representing my actual molecule?
Another thing I noticed is that the 3D geometry of the molecule is not correctly represented. Marvin draws the Pt complexes in a tetrahedral instead of square planar geometry. Does this affect the log P prediction?
I would propose to do these steps
1. delete the dative bonds from the structure
2. calculate the logP (2.85)
3. add a "ring formation" increment to the logP (2.85 + 2 times 0.3 = 3.45)
The "0.3 logP unit increase/ new ring" is an empirical value.
See the attached figure.
The 3D geometry is really bad. Luckily it is not taken into account in the logP calculation.
I hope it will be fixed by the "3D cleaner" developers.
I've tried that for one of the molecules I'm working on (Zr-DFO) and compared it to the literature logP value:
Drawing only the three hydroxy Zr-O bonds and leaving the three dative =O-Zr bonds away, I got a clogP of -0.67.
If I add the empirical value for the three ring fomations this gives -0.67 + 3*0.3 = 0.23
The experimental value for this Zr-DFO complex, however, was reported as -2.83 (Pandya et al. 2015).
So I guess for this kind of complexes the calculations don't really work. Or is there another workaround that should be used?
Here's the literature reference mentioned above: Pandya et al. Chem. Commun. 2015, 51(12), 2301-2303.
The Zr chelator is desferrioxamine B (DFO). The DFO has an ionizable amine group.
The amino group will be ionized at the experimental pH value, and this is why the logP will be decreased.
Look at the attached figure.
For example : The "logP" value of the modified "Zr -DFO" is -3.68 at pH = 7.
The "logP" of the "Zr -DFO" = -3.68 + 3x0.3 = -2.78
If possible please check the experimental pH value and recalculate the "logP" value according to the logD curve.
Oh obviously, thank you for bringing my attention to this. With the charged amino group the value matches the experimental data much better.