JChem Base generates different calc derivatives from Accord

User 8e3c183078

13-05-2013 18:37:01

Hi,


 I am using JChemBase to generate the following in the database columns. The problem is that the calculated values are very different from the Accord software that we used to use before we switched to JChem.


 The Accord software shows the following: (note: aacn is an internal drug number column for the drugs attached in the file (drugs.sdf) here.








































































































accn



intAcceptorCount



intRotatableBondCount



intDonorCount



strLogp



2



5



15



3



3.75



14



5



8



0



1.22



21



5



2



3



2.34



22



5



11



4



3.37



23



22



51



19



-5.93



28



2



10



0



5.51



32



1



0



0



3.99



34



1



3



1



4.26



36



3



3



1



5.59



38



5



2



0



3.30



53



4



6



3



2.36



56



7



9



0



2.91



60



8



5



1



3.87



 JChemBase is generating the following data points for the (same) corresponding Accessions numbers:








































































































accn



intAcceptorCount



intRotatableBondCount



intDonorCount



strLogp



2



5



12



3



3.51



14



5



8



0



1.88



21



4



2



2



2.94



22



5



7



4



3.54



23



20



40



17



-5.23



28



3



10



0



5.49



32



2



0



0



3.83



34



1



1



1



4.42



36



3



2



1



4.38



38



2



3



0



3.41



53



4



5



2



2.42



56



4



9



0



3.63



60



6



4



1



3.2



 


You will see that there is a variation for those highlighted in yellow in what JChem Base has generated as compared to what Accord has generated.


 Is there any thing we can do to get these values in line with what our previous software was generating?


 Thanks, Priya


P.S : I am attaching drugs.sdf file here, that has drug information that lines with the accno column.


 

ChemAxon 9c0afc9aaf

14-05-2013 05:33:24

I have moved this topic to the section for calculations.


Please let us know 


- the JChem version used


- how he values were calculated (if with chemical terms, then the expression, otherwise any parameters that were set )


My colleagues will respond soon.


Best regards,


Szilard

ChemAxon 049d8dcc29

15-05-2013 18:49:58

Dear Priya,



    Thank you for your post.



    I have looked into some of the examples. I believe the value
difference can be caused by different interpretation of scientific
terms. Although I cannot comment on how Accord calculate these
values, let me explain how ChemAxon calculate these terms.



1) For rotatable bonds, ChemAxon identifies these bonds
non-rotatable.

- non-single bonds

- bonds to hydrogens

- bonds to terminal atoms (CH3, NH2, OH etc)

- ring bonds

- bonds of amides, thioamides, suplphonamides

- bonds connecting aromatic rings having 3 ortho substituents

Therefore, molecule with accession #2 contains 12 rotatable bonds.



2) ChemAxon calculates logP using a modified version of the method
described in this paper. I cannot comment on Accord's algorithm. However, although the
values between Accord and ChemAxon logP calculations are different for the same molecule, they are considered
within error margin.



3) For hydrogen bond and acceptor calculation, take molecule with
accession #21 as an example, the donors and acceptors are shown in
the attached image. In Accord's calculation, the nitrogen
which is part of an amide group can be considered as a hydrogen
bond acceptor if you do not consider the delocalization in the
amide group.


I hope this answered your question. If you have any other comments,
please feel free to reply to this post.



Thank you for your support

David