Fast Protein-Protein Docking Algorithms

    Since many proteins bring out their biological functions by binding to a specific partner protein at a specific site, predicting and determining the structure of a given complex is one of the most important focuses for molecular biology researchers. This is often called "the protein-protein docking problem". It is not only an important task for understanding the complex interaction network inside the cell but also the key to rational drug design. Of the many docking approaches, grid-based Fast Fourier Transforms (FFT) has been shown to provide the best balance between complexity and accuracy of computation.

    The Protein-Protein docking problem has fascinated by biophysical chemists and computational biologists the late 1970s. Given three dimensional (3D) structures of two interacting proteins, a docking algorithm aims to determine the 3D structures of the complex. But about this Project and open questions :

    The first block of questions arises from the modeling of the molecules. So far we used a so-called double skin model to represent the molecules. There every atom gets weightes according to whether it is in the skin or in the core of the molecules. And the next point, is that around each atom kernel we put a 3D Gaussian kernel to describe the electron density. The questions that arise, is the Gaussian really the most suitable function to describe this? Or can we improve the model by using the Lennard-Jones potential or other functions?

    The second block of questions arises from the actual computation of the molecule complexes.

In the docking procedure there comes a point when we need to search in the space of 3D motion, how the molecules fit together best? 

So far we could either search fast over the space of translations and the space of rotation. A question might be whether we can find a unified approach for all motion which would lead to a Fourier transform (FT) on the Motion group SE(3). 

-  But also the approach of fast rotational docking needs more investigation and is not ready implemented. Particularly, we should solve, How we can compute sufficiently fast and exact the infinite sums appearing; How good the results are; What approximation errors occur and where; How the computation can be improved; How much time the computation takes; How the results compare to algorithms from the other groups; ... .



People in this Project

Sajdeh Sajjadi, M.Sc.Computing in Structural and Cell Biology


No publications so far.