Computer Simulations of Protein Folding

By Leili Javidpour

NOTE: This is an overview of the entire article, which appeared in the March/April 2012 issue of the Computing in Science and Engineering magazine.
Click here to read the entire article.

Computer simulations have helped scientists understand the fundamental features of protein folding. In fact, much of our present knowledge about protein folding has been discovered or verified by computer simulations. We greatly rely on these simulations, because other experimental apparatuses can’t reveal protein dynamics with enough spatial and time-interval resolution. Additionally, computer simulations act as a flexible laboratory to test theoretical models, or even inspire with their results, so that scientists are able to develop more realistic theories for different aspects of protein folding.

However, there are many ways that computer simulations still seem “limited” to scientists, and they aren’t always the best option for finding out specific information about the protein-folding process. In this article, the author provides some background information on proteins and what is known about the protein-folding process. She then discusses the role (and limitations) of computer simulations in discovering or verifying information regarding protein-folding theories, and consider when it’s best to use computer simulations versus other scientific resources and approaches.

Protein Folding

Proteins are a large class of biomolecules that exist in nature. They’re involved in nearly every function and mechanism happening in living cells. They even contribute to the production of new protein molecules. These biopolymers consist of 20 different amino acids as their building blocks.

Proteins are biologically active in a special molecular configuration, with a special geometrical shape and exposed active parts that are the protein’s so-called “folded state” or “native state,” also known as their tertiary structure (see Figure 1). The process in which a protein begins as a nonstructured chain and finally adopts its folded state is called protein folding. Researchers have found the native structure in the atomic resolution of many proteins by using x-ray crystallography and nuclear magnetic resonance (NMR), for example.

Why Study Protein Folding?

The study of protein folding is an important and highly active field of research. It’s important to study this process’s potential problems for many reasons. The most obvious reason is that a misfolded protein wouldn’t have its biological functionality. The misfolding of proteins could happen normally in cells, and in fact there are some mechanisms for recycling misfolded proteins. However, if protein misfoldings happen much more frequently than usual, they could cause dangerous diseases such as Alzheimer’s, Parkinson’s, and prion diseases. On the other hand, if we know how a protein sequence folds, then we can design some proteins to have a special structure and do a special task in living cells. This field is usually referred to as protein engineering. This field actually has a good success rate for smaller proteins.

Protein folding from an unstructured amino acid sequence occurs as a complicated process of forming a special interaction among many different possible interactions on the amino acid chain. Really the phenomenon is a probabilistic one – namely, a protein sequence won’t always fold to the folded state, but it will fold with a probability that depends on the sequence itself, the solution’s pH (that could in turn change the stability of different interactions in proteins), and the solution’s temperature. This is why studying protein folding has become so closely related to its atomistic study.

Computer Simulations

Protein folding, is a very complex process to study, and there are many approaches that can be used in an attempt to learn more about this process. Computer simulations can use many different models of proteins to study their folding. As with any other scientific field, depending on available computational power, different levels of accuracy can be achieved. If the goal is to study a general behavior for many different kinds of proteins, then it’s likely that simulations can use a greatly simplified model of proteins.

The author describes simple latttice models of protein structure, going on to more complex non-lattice models, and much more complex atomistic models. Lattice models are from the simplest models of proteins’ structure. In these models, each amino acid is usually represented by only one particle or bead, and it’s restricted to move only on a specific lattice site. In a more realistic class of non-lattice models, beads aren’t restricted to a lattice. The article then addresses the interactions among beads in various models.

Computer Simulation Methods and Analysis

The author discusses at length the methods being employed to simulate proteins in computer analysis. Molecular Dynamics (MD) and Monte Carlo (MC) methods are the two main classes for the simulation of proteins, similar to other molecular systems. An MD simulation numerically integrates Newtonian equations of motion for all particles that are defined in the protein structure model.

What Results Can Simulations Provide?

Simulations are in fact a laboratory to verify our theories or even to help develop new insights. Although the time and spatial resolution in a computer simulation is great enough, many different system-model properties can be measured. The author mentions some of the most common measurements that are usually the most helpful in analyzing protein-folding dynamics, including the folding and unfolding rates, and their free-energy surface.

Many groups have studied the protein-folding problem. Using computer simulations to study the problem is a great help, except for the present shortcomings with regards to limitations of time and space resolution. However, by using the high computational power of computer clusters that are being developed in more institutes every year, over time the community will be able to extend our knowledge and ability to predict the folded state of an amino acid se- quence to larger and more complex proteins.

ABOUT THE AUTHOR

Leili Javidpour is a postdoctoral associate at the Institute for Research in Fundamental Sciences. Her research interest is in computational biophysics. Javidpour has a PhD in physics from the Sharif University of Technology. Contact her at javidpour@ipm.ir or javidpour@alum.sharif.edu.