ACD/pK_a DB Algorithm Training

Since prediction is most valuable for novel compounds and classes, for which no measured values may be available, it is essential to have a means of expanding accurate prediction to these novel chemical classes. ACD/pK_a DB contains two tools that enable such customization:

System training

This tool allows you to train the algorithm with experimental pK_a data that can be taken directly from the user's database of experimental measurements. It's applicability is restricted to larger compounds that contain the macro-substructure of the compounds used for training. It is valuable, for example, in a situation where the scientist is working with analogs of a parent compound with minimal changes in structure.

ACD/pK_a Accuracy Extender

This extensible, expert tool is used to design custom Hammett/Taft equations for novel chemical classes. Resulting equations can be used for all calculations made for this class by any ACD/Labs product that employs pK_a prediction. It is especially valuable for evaluating properties of novel acid/base heterocycles since accuracy extender allows users to define a new ionization center and point of attachment for a variable substituent, and assign a relevant Hammett/Taft equation to define and improve the accuracy of pK_a prediction for new chemical classes.