From WebChem Wiki
Jump to: navigation, search

Atomic charges, or atomic partial charges, are non-integer numbers quantifying the balance of positive (nuclear) charge and negative (electronic) charge associated with each atom. In the 3D space, atomic charges represent points placed at the position of the atomic nuclei, and may be termed atomic point charges. The molecular representation based on atomic point charges is thus a very basic abstraction of the molecular electron density. Having only a conceptual character, there is no unique definition of atomic charges. Rather, a score of such definitions have been published and are in use.

ACC is able to calculate atomic charges on molecules of any nature and size based on empirical models. What one can do with the resulting atomic charges strongly depends on the principles behind the definition of the atomic charge concept used in the development of the empirical model. Be sure to also check out the Frequently Asked Questions section for additional information.

While atomic charges are merely concepts and not physical observables, they have been used heavily in theoretical and applied chemistry due to their highly intuitive character and correlation with measurable quantities such as the electrostatic potential, polarity, reactivity, etc. Nowadays, atomic charges are still integral parts of many modeling applications, and are still used in reasoning basic chemical processes. Below you can find a few tips regarding the main applications of atomic charges in understanding and modeling the chemical behavior of molecules. This is by no means an exhaustive list. It is useful to keep an open mind with respect to potential applications, while being aware of the limitations inherent to the atomic point charge model.

Chemical reactivity

Many reaction mechanisms rely on a surplus of positive or negative charge at a certain site in the molecule. For instance, a nucleophylic attack happens at a positively charged site. In simple molecules like tert-butyl bromide, such sites are easy to identify by electron induction effects. However, as molecules become more complex, it is difficult to distinguish potential reaction sites based solely on concepts like induction and conjugation, because they are not quantitative. A simple atomic charge calculation can easily identify potential reaction sites.

Such a concept can be extended to biochemically interesting sites. Due to charge transfer and polarizability, total residue charges will deviate from the expected formal values (-1,0,1). Specific residues which are significantly more positive or negative than their surroundings are more likely to be the site of post translational modifications. Some salt bridges are stronger than others, and it is possible to predict which salt bridges can be more easily disrupted.

Critical elements enabling the transition between two states

The absolute values of atomic charges cannot be validated experimentally. Sometimes these values correlate with observable phenomena (potentials, chemical shifts, etc.). However, even when the absolute values of atomic charges do not correlate with any phenomenon observable for a certain molecular system under given conditions available to the user (or in literature), it is still possible that the relative differences in charges hold further information about chemical reactivity and biological significance.

Given two states of a molecular system, patterns in charge differences can provide insight into the mechanism by which the system evolves between these states. The two states can be different conformations, mutants, ligand bound and ligand free states, etc. Look for significant differences in the charge on atoms, residues or other relevant molecular fragments. These sites are most likely important elements of the transition, and acting upon these sites can modulate the ability of the molecular system to move between the two states. Such information can be important in protein engineering, understanding disease, etc.

QSPR and QSAR modelling

The main concept behind Quantitative Structure-Property Relationships (QSPR) is that different molecular structures have different properties, and thus a (generally physico-chemical) property of interest can be represented as a function of descriptors derived from the molecular structure. In Quantitative Structure-Activity Relationships (QSAR) modelling, the property of interest is the molecule's activity in a certain context (e.g., biological activity).

Atomic charges are used as descriptors in various QSPR/QSAR models, generally in combination with other descriptors (based on the atomic composition, molecular topology, 3D structure, electronic structure, etc.). Some properties correlate with atomic charges better, and will be more easily predicted by QSPR models that rely on atomic charges as descriptors. For example, it is possible to predict dissociation constants by QSPR models which use only atomic charge descriptors, whereas predicting properties related to toxicity require several different kinds of descriptors.

Generating molecular conformations

The energy and properties of a molecular system depend on the 3D molecular structure. In order to predict the binding properties of a ligand to a target (e.g., before the synthesis of a potential drug candidate) it is necessary to have information about the possible conformations of the ligand at the given active site. Such information can be obtained by modeling techniques (docking, molecular dynamics, etc.). However, reaching a suitable ligand conformation from an initial state which is very different from a conformation that binds successfully can be very demanding computationally. To ensure that plausible conformations are sampled, it is often useful to start with several different conformations of the ligand.

Many tools have been developed for the generation of different molecular conformers. These tools employ various heuristics for the generation of thousands of conformations for a single ligand molecule, and then evaluate the energy of each conformation. Atomic charges are often used in the estimation of the electrostatic contribution to the total energy of each particular conformation.


Molecular docking attempts to find suitable orientations between two molecules that form a complex. In the first stage, a set of possible orientations are generated. For example, the receptor is kept in a fixed conformation, while thousands of conformers are generated for the ligand using various algorithms. Then, the energy of each possible orientation is estimated, and the orientations are ranked. Atomic charges are often used in the estimation of the electrostatic contribution to the total energy of each particular orientation.

Many docking tools are available, each different and continuously evolving. Currently, ACC does not provide input files specific to each docking package. However, ACC does provide molecular structure files containing charges (mol2, pqr), as well as files containing only charges (.mchrg) which can be easily incorporated in other file formats.

Molecular Dynamics simulations

Molecular dynamics uses classical mechanics to model molecular systems. The energy of a molecular system is calculated using force fields. Force fields express the energy of the molecular system as a sum of energy terms (e.g., for bonded and non-bonded interactions), each expressed as a function of certain parameters. Atomic charges are used in the estimation of the electrostatic contribution to the total energy of the molecular system.

In many force fields the parameters may have been fitted to work optimally with a certain kind of internal atomic charges (e.g., AMBER uses RESP charges, CHARMM uses CHARMM-like charges, etc.). Mixing external charges into such a force field may thus not give an optimal behavior of the force field. Unpredictable results may be obtained when employing charges which are fundamentally different from those originally involved in the development of the force field. Secondary structure elements might deviate from their optimal form (bent or partly unwound alpha helices). On the other hand, this is a very effective way to sample conformations of the molecule otherwise inaccessible to the simulation. You may then minimize each new conformation, and run separate classical simulations. This can be a great advantage when simulating processes that involve large conformational changes (e.g., signal transduction via membrane proteins).

Note that many widely used force fields consider a fixed charge model, meaning that atomic charges remain unchanged throughout the entire simulation. By nature, the empirical model used by ACC produces atomic charges which are specific to the 3D structure, and thus respond to changes in molecular conformation or chemical environment (e.g., movement of ions closer to the binding site). This means that if you plan to use a force field with a fixed charge model but you still want to allow the electrostatic environment to evolve during the simulation, you will have to update the atomic charges manually at certain intervals during the simulation (e.g., by running ACC on snapshots of the molecular structure, and then continuing the simulation with the updated charges). While such a procedure is easy to implement, always remember to proceed with care when mixing charges which are external to the force field, with the rest of the force field that has been developed together with a set of internal charges.

Many molecular dynamics simulation tools are available. Currently, ACC does not provide input files specific to each simulation package, mainly because there are too many of them, each different and continuously evolving. ACC does provides molecular structure files containing charges (mol2, pqr), as well as files containing only charges (.mchrg) which can be easily incorporated in other file formats.

However, if you are primarily interested in running molecular dynamics, we recommend to look into ReaxFF, a force field which works with atomic charges that respond to changes in conformation and chemical environment. The principles of atomic charge calculation in ReaxFF are similar to those applied by ACC. Efficient ReaxFF implementations can be found in LAMMPS or the commercial tool SCM.

Start by having a look at the main terms used by ACC, or return to the Table of contents.