Proteomics/Protein - Protein Interactions/Protein Space

Page Edited and Updated by: Dan Surdyk
E-mail: dfs6389@rit.edu

This Section:

Introduction

Protein-protein interactions are extremely dependent upon the three-dimensional structure of involved proteins, more-so than just their amino acid sequences. The way a polypeptide chain folds into its operational condition in vivo has long been determined by experimental techniques like X-ray crystallography and NMR. Both of these methods are long, laborious processes that often deliver questionable results, and at best are only elevant to the protein being studied or proteins that are known to be closely related. Recent advances in science and computing have enabled 3D structures of proteins to be accurately predicted for a subset of small, simple proteins using nothing more than their amino acid sequence. As an offshoot of this, protein-protein alignments can be done according to structural features as well as amino acid sequence, suggesting the function of uncharacterized proteins with no known relatives and thus providing insight into the way they will interact with other proteins in a system.

Prediction Methods Using Protein Space

Many prediction methods used to predict three-dimensional protein structure are the same as those used to predict how proteins of known structures will interact (click here). The "correct" way of prediction is to try and score all possible conformations, which is computationally intensive. The only known work-around is to develop short-cut methods that sacrifice 100% accuracy to gain speed. Emerging trends in computational structure prediction combine algorithms with biological knowledge to take a "smart search" approach, only focusing on areas of proteins that are most likely to be biologically significant. This is an admitted improvement over the "random search until a near-optimal configuration is found" approach.

Protein Structure Alignment Using Structural Space

File:Structure space map.jpg

An example of a structure space map.

Another direction that is being taken is to adapt primary sequence alignment methods to consider secondary and tertiary structure. There are a number of features of proteins integral to their function and interaction with other proteins that are not determined by amino acid sequence alone. Many proteins share functional properties despite vast sequence differences because of the shapes that they fold into. One method of simultaneously quantifying and visualizing these relationships is using a protein structure space map. [¹] Roughly speaking, a Protein structure space map is the result of scoring how well known proteins match, structurally. That score is used as a directional distance used to position families of proteins in relation to each other on a set of axes, closer if they are more similar, distantly if they are more dissimilar. Anyone can look at the structure space map and immediately judge how similar two proteins or protein families are by their proximity on the map.

There are a number of structural alignment tools and databases available commercially and freely on the web. A good list of them can be found on the Wikipedia page for alignment software.

Further Reading

• Shindyalov, I. N., Bourne, P. E. Protein Structure Alignment by Incremental Combinatorial Extension (CE) of the Optimal Path. Protein Engineering 1998. 11(9) pp. 739-747.
• Wikipedia page on Structural Alignment