Co-authors Alberto Belmonte Parra, PhD and Christophe Vaillant, PhD
Simulating molecular systems can be a time consuming process, especially when trying to model systems that form complex structures. Unfortunately, these types of complicated systems are common in industrial applications across different sectors, making certain simulations particularly inaccessible to formulation scientists. One way to accelerate these simulations is by exploiting coarse graining methods.
Coarse graining is the act of building a simplified representation of a complex system while keeping the most relevant chemical and physical properties. In molecular dynamics, the coarse graining of a molecule refers to the act of grouping together atoms into units. These units, namely “beads”, are connected by harmonic bonds and constraints on angles to ensure that the structural properties of the molecules are well represented.
In general, coarse-grained models reduce the computational cost of simulations, thus enabling us to simulate longer, larger and more complex systems.
Coarse graining is a complex process of translating a molecular representation into a set of beads and bond definitions that are used to compute the force field during the simulation. However, many force fields, such as MARTINI (see http://cgmartini.nl/), have a complex library of bead definitions and often rely on human input to coarse grain specific molecules. In RheoCube, we use the SDK force field, which has a consistent library of beads and bond definitions in terms of specific chemical groups, and allows us to automate the coarse graining process. For that, we use SMILES (simplified molecular-input line-entry system) strings as an input to represent the molecular structure. Using the open-source tool RDKit, we translate the SMILES string into our set of beads and bonds definitions.
One of the biggest advantages of the automated coarse graining with RheoCube is the possibility to build up a molecule simply from its chemical structure. There is no need to know properties such as density or viscosity. Moreover, it allows us to test the behavior of potentially interesting as-yet unsynthesized molecules before any lab work is carried out.
Chemical interactions between ingredients are highly dependent on their molecular structure. From linear structures to rings and branched structures, capturing all structural details is essential to perform reliable simulations. Here, we present some examples of molecules commonly used in industrial applications which exhibit different complexity in their chemical structures.
Polyoxyethylene oleyl ether surfactants are commonly used as structure-directing agents, for instance in the synthesis of zeolite-based catalysts. In Figure 1, we show the translation from the SMILES string into a set of beads, as well as a 3D representation of the molecule with an overlay of the beads. In this case, we have chosen a rather simple linear surfactant with 20 ethoxylate units. RheoCube was able to correctly coarse grain the structure into a set of beads that properly represents the chemistry of the molecule. We can see the presence of ethoxy beads, single and double bond carbon beads, as well as a hydroxyl bead.
Figure 1. a) Chemical structure of polyoxyethylene oleyl ether with 20 ethoxy repeating units (POE20). b) SMILES string used in RheoCube as an input. c) Beads generated by the coarse-graining method. d) 3D visualization of the coarse-grained molecule showing the different beads and connections.
Many molecules exhibit more complex structures, such as rings and branches. The automated coarse graining method of RheoCube is capable of identifying and translating most of them into a coarse-grained representation. An interesting example is the well-known cholesterol molecule. The chemical structure of cholesterol has a central sterol nucleus with four hydrocarbon rings and a hydroxyl group. In Figure 2, we show the 3D visualization of the coarse-grained molecule after parsing the SMILES as input. We can see that RheoCube succeeded in representing the structure. It is interesting to point out the presence of ILE groups which capture the presence of the pending methyl groups from the hydrocarbon rings.
Figure 2. Chemical structure and 3D visualization of the coarse-graining of cholesterol.
Molecules having aromatic rings, such as benzene rings, are commonly used in pharmaceutics. Two clear examples are aspirin and paracetamol. Both molecules have a benzene ring and other functional groups attached to it. In Figure 3, we show their coarse-grained structures. We observe that in both cases, RheoCube succeeds in creating a proper representation of the molecules. Taking a closer look at the beads composing the benzene ring, we can see that the paracetamol has a phenol-specific ring bead to properly capture the presence of the hydroxyl group.
Our last example includes branched molecules, such as branched polymers. These molecules present very interesting properties due to their loosely packed structure. For example, they have a lower density in comparison to the linear structure, and also exhibit lower melting and boiling points. In Figure 4, we show the coarse graining of a hyperbranched polyester. Branching adds extra complexity to the coarse graining process due to the multiple paths that the algorithm must traverse. In this case, RheoCube succeeds in coarse graining the branched structure in a few minutes of processing.
The implementation of automatic coarse graining in RheoCube allows users to integrate ready-to-use new molecules in their cabinet of ingredients in a single step. By simply typing the string of the molecule, RheoCube can translate that molecule into a set of beads which hold all necessary properties for the simulation. We have demonstrated RheoCube's ability to coarse grain molecules with structures whose complexities range from simple to complex, with the presence of rings and/or branches, thus highlighting the flexibility of RheoCube's algorithms. It must be acknowledged that the list of beads in the SDK force field is limited and some molecules might not be possible to coarse grain. We are actively working on extending the library to different chemistries.