MoKa. New version 2.0
Accurate pKa prediction and automatic structure modification is critical for many computational chemistry methods which are strongly dependent on the tautomerization and protonation state of the structures, including docking, binding affinity estimation, QSAR and ADME modelling, and metabolism prediction.
MoKa implements a novel approach [1] for in-silico computation of pKa values; trained using a very diverse set of more than 25000 pKa values, it provides accurate and fast calculations using an algorithm based on descriptors derived from GRID molecular interaction fields.
Key Features:
- accurate: prediction error of 0.4 log units (0.7 for novel structures)
- fast: more than one million predictions per hour
- automatic: output top N most abundant species at a designated pH
- self-trainable: incorporate additional experimental data to improve accuracy
MoKa 2.0 enhancements over MoKa 1.0
- Some ionizable centres not identified by MoKa 1.0 have now been added (the total number of predicted centres is now 106)
- The number of models describing the chemical space has been expanded from 35 to 53.
- Hundreds of molecules have been synthesized and experimentally tested to improve the training of a number of models
- Some models in MoKa 1.0 were found to display an overfitting effect; this has now been fixed improving these models' external predictivity.
- Tautomer enumeration is now quasi-exhaustive, and also independent of the input tautomer.
- Model training has been improved, highlighting the impact of user data on the built-in models.
- The analysis of experimental pKa versus predicted values has been improved, making it easier to focus on any poorly predicted compounds.
- Redundant protomer enumeration is now allowed as an optional argument, so that both protonation states are produced when there are two ionization states with a close predicted pKa difference
Package content
Graphical interface for predictions
- tautomer check
- batch mode for multi-structure files
- integrated structure editor
- cut & paste from ISIS/Draw (Windows version)
Automatic comparisons with experimental data
- support validation of MoKa predictions
- compare results with other prediction methods
- evaluate benefits of training module
System training module
- independent GUI for system training
- automatic assignment of exp. pKa values to ionization sites
- building of customized pKa prediction models to load into MoKa
Command line tools
MoKa command line
- easy integration with existing tools
Generation of ionization states of a compound
- single ionization state mode at user-specified pH value
- multiple ionization states mode at user-specified pH value or pH range
Tautomers enumeration and stability estimation [2]
- batch enumeration of tautomers and estimation of tautomer stability
Availability
MoKa is available for Windows® and Linux® operating systems.
Editions comparison
MoKa comes in two editions: Solo and Suite
MoKa Solo provides an interactive interface for pKa investigation; Moka Suite provides unlimited interface access and also command-line utilities for wider enterprise deployment in cheminformatics processes.
This matrix provides an overview of what is provided by the different editions.
| Solo Edition | Suite Edition | |
|---|---|---|
| Graphical interface for predictions | yes | yes |
| Automatic comparisons with experimental data | yes | yes |
| Command line batch predictions | no | yes |
| Generation of ionization states of a compound | no | yes |
| Tautomers enumeration and stability estimation | no | yes |
| System training module | no | yes |