General molecular mechanics force fields usually employ a large number of molecules as the training set to obtain corresponding force field parameters, which, in turn, can be applied to new molecules. Good transferability enables them to adequately describe various molecular systems with the same set of force field parameters. However, when it comes to specific systems or properties, these general force fields cannot provide a high level of accuracy required by many calculations, such as crystal structure prediction (CSP). By contrast, customized force fields are better positioned to meet the demand for high accuracy with their parameters optimized explicitly for a given molecular system.
According to specific characteristics and physical forms of a drug molecule, a customized force field constructs its own training set and energy functions. It then adopts corresponding parametrization strategies to more accurately describe the target characteristics of the drug molecule.
Compared to high-level Quantum Mechanics calculations, the accuracy of customized force fields can reach the level of kJ/mol, much better than that of general force fields.
Upon customers' demands, the customized force field is widely applicable in accurately describing specific molecular confirmations, molecule interactions, solid or liquid properties, and more.
XtalPi's customized force field's training process is fully automated, without the need for human intervention, and can be completed within a single day.
The customized force field can be improved iteratively in CSP or specifically in predicting specific molecular properties.
The combination of local and global intelligent optimization algorithms can achieve a fast and efficient optimization even in large parameter space.