I have used JChem for Excel with great success to calculate a range of different physio-chemical properties of compounds. However, for the LogP (and LogD) values I can see that for some compounds they are very close to experimentally determined LogP values and for some they are not. I am therefore wondering if the algorithm used to estimate LogP (and LogD) considers hydrogen bonds that might occur within the molecule in aqueous solutions? E.g. some of the compounds that I am looking at have several ring structures which might interact with each other in aqueous environment.
Anne Marie Ellegaard
Dear Anne Marie,
the logP/logD plugin takes into account H bonds in general, but there might be some cases when it does not handle them correctly. Can you send us an example that failed ?
Thank your for your reply.
The reason why I'm looking into the LogP (and LogD) algorithm is because I'm working with a range of compounds that contain aromatic rings. These could potentially be sheltered from the aqueous environment in the cytosol of a cell by internal hydrogen bonds. In this case I would expect the LogP (and LogD) to be affected.
I have found some experimentally determined LogP values in the literature and compared them to the values calculated by JChem for Excel. For some compounds the two values are very similar, for some not.
Here are some examples of compounds where the values are not very similar (see also attached document):
|Log P (JChem)|
Along the same lines, I'm wondering if it is possible to calculate LogP (and LogD) at different pH-values and at different temperatures. I am interested in the effect of some of these drugs in the lysosome of cells, which has pH 4-5. And as I am working in human cells, I would like the temperature to be 37 degrees. I assume that both pH and temperature would affect the LogP and LogD. Is it possible to alter these things in the algorithm?
thank you for sending the valuable data. We are going to develop our logP prediction model so that the logP for this set will be better. However, we cannot take temperature effects into consideration, as this would require a lot of thermodynamical calculations.
The development of our logP calculation will be available soon.
Thank you for your reply.
I'm still not sure if the algorithm considers the possiblity of internal H bonds within a compound. Could you please clarify this for me?
Thank you in advance.
what makes you worry about the internal H bond handling? Should you send us an exact example, we could give more detailed explanation.
For example , because of the H-bond the logP value is increased with 0.65 in case of salicilic acid. See the attached figure. Choose the VG model on the Marvin sketch and perform logP calc.
The Structural increment = 0.65 and 0.63 for neutral and ionized salicilic acid. This is the effect of the hydrogen bond on the logP value.
It can occur that not all donor-acceptor interction is considered as potential H bond couples.
We are doing development in this direction too.
Dear Daniel and Jozsi,
Thank you for your quick responses and the example of salicylic acid. Though, it does not quite illustrate my point.
A better example could be lidoflazine. The compound contains several ring structures, which might interact with each other in an aqueous phase to protect the hydrophobic rings from the water molecules of the surroundings.
As you can see from the table I provided earlier in this discussion, the LogP value calculated by JChem is 6.58, whereas an experimentally determined LogP is 5.60. I'm wondering if this difference could be due to the fact that such internal H bonds change LogP of lidoflazine in the experimental setting, but are not considered in JChem's algorithm. That is why I'm asking if this type of H bonds are part of JChem's algorithm?
I'm looking forward to you answer. Thank you in advance.
our logP prediction model mostly takes the H bonds into account. Chemical features that cannot be attributed to a single atom (but which can be attributed entirely to the molecule) are taken into account and displayed in the result window as Structural Increment. In case of lidoflazine no such structural increment was detected, therefore the difference between the measured and calculated logP comes from the inaccuracy of our current calculation model, not from not taking the internal H bonds into account. We are planning to develop our logP prediction model in the near future.
However, I would also examine the accuracy of the experimental value; measuring such high logP values is quite difficult, so the accuracy of the experimental data should also be checked.
Thank you for your quick and thorough answer. This was exactly what I was looking for.
Have a nice day.