Performance of JCF.evaluate inside Oracle package

User f3a11790a9

23-05-2014 09:17:09

Dear Support,


We download ACD (RD files) from Accelrys and upload to Oracle schema through a customized package. The package uses JCF functions such a jcf.evaluate and jcf.molconvert.


We have seen performance issues with the usage of these functions and if these functions are commented out, the data gets loaded in native Oracle schema much-much faster.


Please could you suggest some generic guidelines to improve performance while using these functions or could suggest some specific steps/work arounds that could be checked.


We use jcf.evaluate to calculate the logp value and use this value to insert in Oracle schema. jcf.molconvert is used to 'convert to smile'.


Any help is welcome.


Regards


RG

ChemAxon abe887c64e

23-05-2014 14:02:02

Dear Customer,


Unfortunately, these calculating processes are really time-consuming operations. Let us ask a few clarifying questions. What is your purpose with these calculations? Do you store the calculated pKa values together with the molecules? What is the reason why you convert the structures into smiles? How often do you plan to run insert processes?


Regards,


Krisztina

User f3a11790a9

23-05-2014 15:53:49

Thanks for your quick response. below are the details


1) We have a requirement to change the native LogP values of ACD files with the JChem calculated values. We use jcf.evaluate to do this. We are not sure how the LogP is set in the ACD files, is it already done via JChem?


2) We dont use pKa values


3) We need the ability to search based on smiles also.


4) We need to download the ACD RD files and upload these in local Oracle schema (with some conversions). This is intended to take place once in 2 months.


Regards
RG

ChemAxon abe887c64e

26-05-2014 10:08:50

Unfortunately we don't know the origin of the native logp data in the ACD files. With jcf.evaluate you can actually calculate logp data,  then store them in an additional step. An other solution -  to calculate and store the logp (or any other chemical term) value of the structures - would be to use the 'autoCalcCt' parameter during the index creation. This second mode can be quicker, because of parallel processing.  See create index details here.


Example:


CREATE INDEX jcxautocalccttest ON autocalccttest(structure)
INDEXTYPE IS jchem_cc.jc_idxtype
PARAMETERS('sep=! autoCalcCt=numeric(30,15);logp()!autoCalcCt=numeric(10,0);rotatableBondCount(

Using JChem Cartridge, you will be able to run searches in tables containing structures in RD file format even with query structures represented in smiles format.


Regards,


Krisztina