10 Applications of Molecular Dynamics Simulations …
183
etc. Predictive models can be developed using different computational methods such
as pharmacophore modeling [1–4], molecular docking [5–9], and machine learning
methods [10–22]. These computational methods reduce animal model use, cost, and
time while improving safety evaluation and risk assessment of chemicals. The emerging field of computational toxicology predicts and examines toxicity of chemicals
[23, 24]. The following tools are mainstays of computational toxicology prediction:
(1) Databases to store chemical data including chemical properties and toxicity.
(2) Software tools to generate molecular descriptors for chemicals [25].
(3) Programs to run molecular dynamics (MD) simulations.
(4) Algorithms for molecular modeling.
(5) Statistical packages to generate two-dimensional (2D) and three-dimensional
(3D) quantitative structure–activity relationship (QSAR) models [26–28].
(6) Web servers or standalone applications to predict toxicity of chemicals using
pre-built models.
(7) Software tools to visualize prediction models.
Among various algorithms and tools used in computational toxicology, MD simulations are increasing in prevalence. MD simulation is a well-known technique in
other fields including drug design, descriptor generation, structural biology, protein
analysis, identification of hot spot residues, and more. In the field of toxicology, MD
simulation can link structural biology with chemical toxicity information. MD simulation informs physical movement of the atoms or molecules in a molecular system
and the structural changes of the protein in a time-dependent manner. This computational technique also illuminates conformational changes and protein fluctuations
induced by binding of various compounds or chemicals. Different algorithms exist
for MD simulations, addressing various aspects of computational toxicology. This
chapter reviews MD simulations as applied to computational toxicology.
10.2 History of MD Simulations
MD is a 100-year-old technique, but only gained traction with the scientific community during the twentieth century [29]. Table 10.1 contains a brief history of MD
simulation. In the mid-50s, Fermi, Pasta, Ulam, and Tsingou successfully developed
the Monte Carlo simulation method [30]. MD simulations build on these statistical
methods. In 1957, Alder and Wainwright studied the interaction of a hard sphere
using MD simulations. The results revealed many key learnings on simple liquid
behavior [31, 32]. The next milestone was achieved by Rahman in 1964, when the
first realistic liquid argon simulation occurred [33]. Next, in the late 70s, the technique
of MD simulations was further improved by simulating several hundreds of atoms up
to biological systems [34, 35], i.e., immersing the whole protein in solution, embedded the protein in a lipid layer, or macromolecular complexes [36, 37]. Rahman and
Stillinger created the first realistic simulation of liquid water in 1974 [38]. In 1977,
McCammon et al. simulated the first bovine pancreatic trypsin protein inhibitor [34].
183
etc. Predictive models can be developed using different computational methods such
as pharmacophore modeling [1–4], molecular docking [5–9], and machine learning
methods [10–22]. These computational methods reduce animal model use, cost, and
time while improving safety evaluation and risk assessment of chemicals. The emerging field of computational toxicology predicts and examines toxicity of chemicals
[23, 24]. The following tools are mainstays of computational toxicology prediction:
(1) Databases to store chemical data including chemical properties and toxicity.
(2) Software tools to generate molecular descriptors for chemicals [25].
(3) Programs to run molecular dynamics (MD) simulations.
(4) Algorithms for molecular modeling.
(5) Statistical packages to generate two-dimensional (2D) and three-dimensional
(3D) quantitative structure–activity relationship (QSAR) models [26–28].
(6) Web servers or standalone applications to predict toxicity of chemicals using
pre-built models.
(7) Software tools to visualize prediction models.
Among various algorithms and tools used in computational toxicology, MD simulations are increasing in prevalence. MD simulation is a well-known technique in
other fields including drug design, descriptor generation, structural biology, protein
analysis, identification of hot spot residues, and more. In the field of toxicology, MD
simulation can link structural biology with chemical toxicity information. MD simulation informs physical movement of the atoms or molecules in a molecular system
and the structural changes of the protein in a time-dependent manner. This computational technique also illuminates conformational changes and protein fluctuations
induced by binding of various compounds or chemicals. Different algorithms exist
for MD simulations, addressing various aspects of computational toxicology. This
chapter reviews MD simulations as applied to computational toxicology.
10.2 History of MD Simulations
MD is a 100-year-old technique, but only gained traction with the scientific community during the twentieth century [29]. Table 10.1 contains a brief history of MD
simulation. In the mid-50s, Fermi, Pasta, Ulam, and Tsingou successfully developed
the Monte Carlo simulation method [30]. MD simulations build on these statistical
methods. In 1957, Alder and Wainwright studied the interaction of a hard sphere
using MD simulations. The results revealed many key learnings on simple liquid
behavior [31, 32]. The next milestone was achieved by Rahman in 1964, when the
first realistic liquid argon simulation occurred [33]. Next, in the late 70s, the technique
of MD simulations was further improved by simulating several hundreds of atoms up
to biological systems [34, 35], i.e., immersing the whole protein in solution, embedded the protein in a lipid layer, or macromolecular complexes [36, 37]. Rahman and
Stillinger created the first realistic simulation of liquid water in 1974 [38]. In 1977,
McCammon et al. simulated the first bovine pancreatic trypsin protein inhibitor [34].