A growing technique in protein structure determination is cryo-electron microscopy. Cryo-EM provides low resolution electron density maps ( ~8 Ångström) . As this data become more accurate and you can use this data not only to determine the overall structure of protein complexes, but also to determine secondary structure elements and their assembly in the tertiary structure of a protein.
The successful field of ab initio protein structure prediction can help to produce possible structures for a certain sequence with unknown fold. To separate structures fitting the electron density map from structures which are different to the electron density picture, advanced fitting techniques are necessary.
The idea is to develop a fitting technique using geometric hashing for fast fitting of multiple modeled protein structure to the electron density map and filter good structures. There are several steps:
1. Transform the electron density map in a point cloud representation.
2. Calculate a hashmap using triangular bases of the pointcloud.
3. Choose atoms to be fit in the target.
4. Search the hashmap and use he basis transformations for the fit.
5. Calculate the standard deviation between the experimental and a simulated electron density map.
Alumni Project Members: Nils Woetzel, Kevin Pereira, Steffen Lindert