JavaScript Tree Menu

Structural Computational Biology

Proteins are the building blocks of life. Understanding them means to understand how living beings work, how diseases work, and potentially how they can be cured. The central dogma of molecular biology looks like this:
DNA ⇒ RNA ⇒ Protein
DNA is transcribed to RNA. RNA is translated into a sequence of amino acids. Our view of this problem focuses on how the three-dimensional structure of a protein forms from a sequence of amino acids. We ignore the process of transcription from DNA to RNA and of translation from RNA to protein. We view the problem like this:
Sequence ⇒ Structure ⇒ Motion ⇒ Function
DNA sequence determines the sequence of amino acids (primary structure) which in turn determines the overall (tertiary) structure; tertiary structure determines flexibility; flexibility entails motion; and motion results in function. Knowing function really is the goal here. My work is motivated by the conviction that robotics can make significant contributions to the first two steps: Sequence ⇒ Structure ⇒ Motion. The last step is best left to biologists.

We are collaborating with Lila Gierasch from the Biochemistry and Molecular Biology Department.

Protein Structure Prediction

Sequence ⇒ Structure
In protein structure prediction one attempts to predict the three-dimensional structure of a protein based on the primary structure, the sequence of amino acids. The structure ultimately determines the function of a protein—that's why we are interested in it.

The motion planning methods for robots developed in our lab specifically address robots with many degrees of freedom. A molecule can also be modeled as a robot which can twist and bend. Large molecules, such as proteins, have thousands of such joints. We are investigating if our methods can successfully be applied to "molecular robots".

In one of our projects we are applying search techniques motivated by concepts from robot motion planning to protein structure prediction. Currently, Rosetta, originally developed in Professor David Baker's laboratory, is considered to be the most accurate protein prediction software. Below we compare the results we have gotten with our method for protein 2PTL, which relies on Rosetta to perform energy calculations for protein conformations, with Rosetta's results and the native structure. This protein consists of 60 amino acids and thus has 120 degrees of freedom. (2PTL really consists of 77 amino acids, but an irrelevant tail has been cut off for the experiments.)


native structure

predicted by Rosetta

predicted using our method

Protein Docking

Structure ⇒ Motion
A protein is a string of amino acids. The arrangement of this string in space is called the structure of the protein. In this arrangement some parts of the protein have become rigid due to interactions between amino acids. Other parts, however, remain flexible. The active site of a protein—the site relevant to the function—very often is flexible. This means that to fully understand to function, one has to understand how the flexible regions of a protein can move. The motion determines how ligands dock to the protein to enable, disable, or alter its function.

By applying methods from robotics we are developing novel methods to determine the motion of proteins and to understand how those motion may result in the docking of other proteins or ligands.