Molecular Modeling: A Detailed Guide

Molecular modeling (MM) and simulation is one of the fastest growing fields in science, and has undergone incredible advancements over the course of the last two decades – particularly as innovations in state-of-the-art hardware and software have led to more practical applications and successes within the field. Today, molecular modeling is an integral part of many virus simulations and drug applications, including very recently, as part of the exploration and investigation of small molecules as potential SARS-CoV-2 drug treatment candidates.

Let’s take a closer look at what molecular modeling is, how it works, and how it’s accomplished.

What is molecular modeling?

Molecular modeling includes all methods, including theoretical and computational, used for building, visualizing, mimicking, or otherwise simulating the behavior of molecules in chemical and biological systems. Molecular models can be built or visualized in three dimensions (3D) or via complex computer simulations on large proteins and nanostructures. 

Fields of study that commonly use MM include: computational chemistry, computational biology, drug design, biomaterials, and spectroscopy.

Why use computational molecular modeling?

As a computer-based technique, molecular modeling is a cost-effective and efficient way to study molecular properties, make predictions, explore novel systems, determine mechanisms and functions, and even design new molecules, processes and products. MM helps researchers and scientists quantify the impact of physical and chemical interactions in silico. In modern drug discovery, computer-aided drug design (CADD) has become a well-respected alternative for high-throughput screening, as it can hasten hit identification, quicken hit to lead, and improve overall lead optimization (including binding affinity, ADMET and more).

What are computer-aided drug design (CADD) methodologies?

CADD is a modern computational technique, which includes molecular modeling, used in the drug discovery process to virtually screen millions of drug molecules in compound libraries into smaller groupings of predicted active compounds. Doing so helps optimize drug targets into lead compounds. The CADD approach has established itself as a transformative technology in the drug discovery pipeline, used not only to accelerate drug discovery, but also to lower its associated costs.

How is computational molecular modeling used in drug design?

With the growth of computational capabilities over the last decade, molecular modeling has been called a game changer in drug discovery research. Some of the most commonly applied computational modeling techniques used in drug discovery include: molecular docking, molecular dynamics (MD) simulation and ADMET modeling.

• Molecular docking - this technique predicts the preferred orientation of one molecule to another in molecular interactions that hold a protein and ligand together in a bound state. Molecular docking is sometimes described as an attempt to find the “best-fit” orientation of a ligand binding with a specific protein. To determine the preferred orientation, docking approaches may include modeling shape complementarity (geometric fit) and docking simulations. Docking simulations are powerful tools that allow for straightforward inclusion of molecular flexibility.

• Molecular dynamics (MD) simulation - this technique is used to analyze the physical movements of atoms and molecules in order to understand the systems’ dynamic evolution, or how the molecules will move over time. MD simulations can illustrate biomolecule and protein behavior in high resolution and with full atomic detail. In a classical MD simulation, forces responsible for a system’s dynamic evolution are calculated using molecular mechanics force fields. There are multiple types of such force fields available, which typically involve quantum mechanical calculations to parameterize. MD simulations can provide information about structure and dynamics, perturbations, and dynamic processes.
  
  
• ADMET modeling - this screening technique is used in pharmaceutical R&D to determine the molecular features that influence a drug’s absorption, distribution, metabolism, excretion and toxicity (ADMET) and identify patterns in the data. A drug’s ADMET profile is a critical factor in a drug’s ability to pass through clinical trials. Today, expansive, easy-to-use, web-based platforms for ADMET prediction have been developed. These platforms typically allow for the creation of high-quality QSAR/QSPR models to predict the activity of new chemicals, in addition to the option of advanced data mining, clustering, and matched molecular pair analysis. 

What are some of the goals of molecular modeling techniques in drug design?

Molecular docking and molecular dynamics can help determine the efficacy of a drug, while ADMET modeling helps predict the clinical success of drugs in clinical trials. These techniques have proven to be critically important in supporting the identification of leads for both experimental in vitro and in vivo testing. Because the pharmaceutical industry struggles with high R&D failure rates for drug candidates, high-performance computing which enables virtual experimentation via molecular modeling and prediction platforms can help reduce expensive late-stage failures for the industry.

For example, in some design projects, researchers are looking for a ligand that both binds to the target and achieves a particular signaling profile. Achieving a specific signaling profile means the drug has stabilized specific conformational states of the receptor and the binding pocket. In this case, designing a ligand requires an understanding of how minor conformational changes in the binding pocket can cause different signaling profiles. Using a molecular modeling simulation like MD can help determine this information.

How is hardware and software for molecular modeling changing?

Progress continues to be made in developing software that enhances sampling, reduces associated costs of computational modeling, and integrates information from artificial intelligence and machine learning methods. Improvements to novel hardware (i.e. graphics processing units and coupled processors) have also been important in the field. Software packages for performing MM and MD simulations have also become more intuitive, making them more accessible to non-experts.

Cloud computing which offers on-demand, real-time computational infrastructure, platforms, and services is having a big impact on computational drug discovery. For pharmaceutical companies, on-demand availability of computational infrastructure (like Platforms-as-a-Service) eliminates the burden to manage an in-house stack, reducing infrastructure costs, accelerating processes and improving scalability.

Virtual resources, like virtual machines, are also on the rise, allowing the management of environments with no direct connection to a host computer. An example of this is the widely used ChEMBL database, a manually curated database which makes its chemical, bioactivity and genomic data and tools available as a virtual machine.