User 406a65392f
10-07-2013 20:26:44
Hi,
I am trying to train the ChemAxon pKa calculation using own dataset. Is it possible to “turn off” the ChemAxon’s dataset and build a trained pKa model using our own data set only?
Thanks!
User 406a65392f
10-07-2013 20:26:44
Hi,
I am trying to train the ChemAxon pKa calculation using own dataset. Is it possible to “turn off” the ChemAxon’s dataset and build a trained pKa model using our own data set only?
Thanks!
ChemAxon d7323a7c96
11-07-2013 11:55:17
Dear Yipin,
Unfortunately, in general we do not allow global pka model building as it needs very many datapoints. How many datapoint would you like to involve in your own model?
According to our experiences, models built with a training set not large enough gives inadequote predictions.
Best regards,
Timea
User 851ac690a0
24-07-2013 09:36:41
Hi Yipin,
I arraived from holiday now and after reading this topic let me to give a supplementary answer for your question.
Our pKa the training can be imaging in the following way:
1. Chemaxon's pKa estimation is the "priori" starting
point of the training model.
2. This "priori" pKa estimation will be corrected
with the user's pKa values.
3.In this sense the predicted pKa value of the "posteriori"
user's model is a "DpKa shift" of the Chemaxon's priori
pKa predictiion.
I just encourage you to use your own training data.If you have any strange experience with the "training" than please share it with us.
Thanks in advance.
Jozsi