Computational Chemistry and Biology
In a nutshell
Modeling reality to understand, predict miniscule complexity
Suppose you want to predict how high a ball dropped from a height will bounce. It depends on the type of ball and the surface it lands on. What if it is dropped into a puddle? What if the ball is magnetic and the surface partly iron? What if the surface changes shape at the ball’s approach?
You might write mathematical formulas to represent the effects of all those factors, combining laws of physics (gravity, magnetism, etc.) and experimentally measured bounce heights.
Then you could program a computer to predict the bounce heights of different balls under varied conditions. Not only is this far faster and cheaper than buying and dropping myriad balls, but your computer models of reality help you understand why a given ball bounces as it does and thus how you could alter one to make it bounce a specific way off a particular surface.
Instead, suppose you want to predict if a potential drug molecule will bind to a target molecule involved in cancer or can pass through the blood-brain barrier to treat Alzheimer’s disease.
Computational biochemists essentially shrink those bouncing balls to the size of atoms (trillions could fit in the period at the end of this sentence) then connect them, in molecular structures, to hundreds of others via spring-like chemical bonds that twist, swivel, and stretch. They place them in water (millions of H20 molecules) as they are in our bodies.
They create computer models to predict how those flexible structures will interact with other molecules and membranes based on factors such as geometries and electromagnetic interactions. Such simulations allow scientists to dissect those interactions and improve drug design.
This modeling is also visualized—rendered as interactive 3D simulations and animations with color-coded details—allowing for atomic-level structural analysis and comparisons as well as for viewing how molecules fit together and interact as parts of vital cellular machinery.
Computational chemists develop and apply computer programs to answer key questions in biochemistry. They model, predict, visualize, and analyze the structures, functions, and interactions of biologically important molecules.
In structure-based drug discovery and rational design, researchers seek to find and create small molecules (ligands) that selectively bind to proteins involved in disease processes to therapeutically alter their activity.
Bioinformatics, visualization, and docking
Computation enables this effort to be faster, more efficient, and less expensive, starting with the search for biologically significant targets. Researchers use bioinformatics—the capability of computers to store, organize, and retrieve vast quantities of data—to map networks of protein interactions, discerning biochemical pathways amid complexity and determining key nodes for intervention.
When a target protein is discovered, computers help predict and refine details of its 3D structure (homology and loop modeling) as well as allowing researchers to visualize and interact with it, alone and in complexes, using programs pioneered and advanced by department scientists (e.g., UCSF Chimera).
Then molecular docking programs—also pioneered and advanced by department scientists (e.g., UCSF DOCK)—can virtually screen and rank potential ligands for the relative strength with which they bind to the targets.
Computer and “wet” complementary
Such programs guide the selection or synthesis of compounds for real-world testing (e.g., in the department’s Small Molecule Discovery Center (SMDC)). Indeed, computer (in silico) and “wet” experiments are complementary: While the former suggests lab-testable hypotheses (e.g., potential drug leads), the latter’s results can iteratively improve computer programs’ predictive accuracy.
In developing and refining their algorithms and mathematical functions, computational chemists combine empirical knowledge with biophysical laws and theory. A docking program, for example, might use knowledge about molecular shape complementarity, laws governing certain atomic-level interactions, and the theoretical effects of water on protein-ligand binding.
Simulation spans gaps
Simulation by computer programs helps to span gaps in empirical knowledge, including those due to the extremely small size and brief time scales of intra- and inter-molecular shape-shifting and interactions.
Such simulations can provide atomic-level details of molecular change over billionths of a second (nanoseconds) or less—levels of dynamic resolution that are difficult to attain by existing physical methods. These models can thus suggest the specific mechanisms underlying ligand-binding effects toward optimization of potential therapies.
Challenges include
- Modeling protein regulation via allostery and post-translational modifications
- Visualizing and integrating bioinformatics and biomolecular data
- Modeling membrane permeation to optimize pharmacokinetics
- Determining enzyme function by predicting substrate specificity