I read your step by step instructions on how to train pKa's and they all seemed reasonable. Unfortunately, when I ran the command:
cxcalc -T pka -t pka1 pka_train.sdf
the command always runs for 1-2 seconds, completes without any error or warning messages, but the pKaReg directory, although created, is always empty (no files; no data).
Is there anything obvious that I'm doing wrong? I'm attaching the SDF file used in the above calculation, plus a second file with protons stripped (also fails).
Note that the non-training command
cxcalc pka pka_train.sdf
Please post here the output of
Here's the output from the windows box that I tried:
java version "1.4.2_03"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.4.2_03-b02)
Java HotSpot(TM) Client VM (build 1.4.2_03-b02, mixed mode)
I also tried it on a linux box:
kumgmw02 7% java -version
java version "1.4.2"
gij (GNU libgcj) version 4.1.2 20070626 (Red Hat 4.1.2-14)
Copyright (C) 2006 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
I'll try installing Sun Java on the latter, but I would like to know your suggestions on the Windows box.
Please see our System Requirements:
- Java distributed by Sun (or Apple's Mac OS X built-in Java)
- Java version 1.5
- For 64bit version of MS-Windows systems, we recommend the 32bit version of Java, since currently Marvin Object copy and IUPAC inchi import/export are supported only in 32bit systems.
You should install a newer JRE (1.5, or above) on your Windows box.
Thank you for the quick response Zsolt!
I installed newer versions of java:
C:\Program Files\ChemAxon\MarvinBeans\bin\pKaReg>java -version
java version "1.5.0_22"
Java(TM) 2 Runtime Environment, Standard Edition (build 1.5.0_22-b03)
Java HotSpot(TM) Client VM (build 1.5.0_22-b03, mixed mode, sharing)
kumgms02 11% java -version
java version "1.6.0_17"
Java(TM) SE Runtime Environment (build 1.6.0_17-b04)
Java HotSpot(TM) Server VM (build 14.3-b01, mixed mode)
But the problem persists on both the Windows and linux boxes.
Note that on the Linux box, I used the JRE packaged with Marvin for developers. I added the location of the new JRE to $PATH in order to stop it from looking to the Gnu Java first. On Windows I didn't do any reconfiguration -- just let the wizard do its stsuff.
I tried installing on a third machine (another Windows machine, but newer). I completely wiped out my old version of java (all components) and installed the newest version from Sun. I then grabbed the Marvin distribution with java bundled (taking no chances) and installed that. This gives:
java version "1.6.0_18"
Java(TM) SE Runtime Environment (build 1.6.0_18-b07)
Java HotSpot(TM) Client VM (build 16.0-b13, mixed mode, sharing)
However the result at this point is the same -- cxcalc runs for 2-3 seconds, but when I look inside pKaReg it's empty.
I notice that a few people have downloaded my sdf files -- if anyone has been able to generate a pKaReg folder based on those structures, could you be so kind as to tar or zip it up and send it to me?
I also checked your file and I was not able to generate the correction library.
The problem is that only five molecules are included in the "sdf" file. This is not enough for the teaching method,therefore, an empty pkaReg folder generated.
You would need add more "phenolic acidic" type of molecules into the "sdf" file. This may solve your problem.
I am working on a more convenient solutionn now.
Thanks, that's very helpful!
Do you know how many we should aim for in order to achieve a satisfactory basis, or is that dependent on the goodness of fit that one is able to achieve with the added molecules?
I have added 8 more phenolic acids to give a total of 13 (see attached), but the training continues to fail.
I have checked your new "sdf" file. It containes some bugs. Please fix the problems and try again to generate the pKaReg directory.
On the attached figures I depicted some problems.
- nitro groups are misstyped
- "pKa" & "ID" couples are specified badly many times.
The atomic index "ID" value always have to be referenced to one of the "phenolic OH" group in the molecule. Since its pKa value is given in the "pKa" field.
- sdf file is corrupted.
I highlited the problematic region in your "sdf" file. The fixed "sdf" file is shown next to the bad one.
Thanks Joszi, and sorry for wasting your time with the last file.
I've corrected the file and checked it over, but it's still not generating a pKaReg -- I attached both an unprotonated and a protonated form.
I appreciate your time thus far, but if you have the chance to take one more look at it I would be grateful.
Yes, really there is a problem. I fixed it now. The fixed version (5.3.2) will be available in this month.
super -- thank you very much for your help!