Chemistry + Computers = Chemoinformatics

by Gabrielle DeMarco on January 18, 2010

Some ingenous chemists here at RPI are combining chemistry with advanced computation to rapidly compare different protein bindings sites. Their results are both novel and beautiful, earning them the cover of the Dec. 28, 2009 issue of the journal Chemical Information and Modeling.

The interactions between different proteins in our bodies can spell the difference between health and illness. These chains of amino acids keep our cells from turning to jelly, keep our bodies metabolizing food, and regenerate a constant supply of new, healthy cells as old ones die. They are also increasingly being used in medications to correct protein dysfunctions within the body. With such a robust resume, you would think we would already know all there is to know about most of the major proteins in our bodies. But, alas, proteins are also among the most complex molecules in nature. They are constantly folding, resulting in a seemingly completely rearranged structure often within seconds.

Their shape-shifting ways can make it very difficult to predict what their function is at any given time. One of the most best way to figure out a protein’s function is to determine what it binds to. Following a chain of protein binding events gives scientists a road map of a specific function without our bodies. But again, proteins prove vexing as the binding sights on each protein surface are often very difficult to compare to each other because of all the folding and movement on the protein surface. How can anyone expect to find a matching binding site when the pesky polypeptides won’t sit still?!

Professor Curt Breneman and graduate students Sourav Das and Arshad Kokardekar’s article within the publication outlines a computational method that does not rely on the specific molecular composition of the binding sites to determine what the protein it is likely to bind to. As they note in the paper, these aspects of the protein can appear significantly different, but still be drawn to the same ligand (think of these as the glue that draws and hold proteins together at their similar binding sites). Instead, they look at the distribution of shapes as well as the chemical environment to map out of the structure of a binding site and then to quickly compare that to other binding sites. They tested out their new method and found it to be extremely powerful to picking binding sites with identical functions. Check out the image below. One of these binding sites is a lot like the other…

And for those of you who enjoy the bigger picture, these results are extremely important to scientists and doctors trying to develop new drugs using proteins. With more information on what these proteins will interact with in the body, researchers can limit some of the dangerous or uncomfortable side effects that have riddled many protein-based drug trails.

The research is all part of the work coming out of the Rensselaer Exploratory Center for Cheminformatics Research led by Breneman.