内容简介：Protein structure prediction is also known as the protein folding problem that attracts the scientists from many disciplines. Typically, the structure of a protein is predicted if its amino acid sequence is given. However, current structure prediction methods are still far from being mature. We have been investigating the problem of combining the structure prediction methods with geometrical constraints that can be obtained from the low resolution 3-dimensional image of protein density. We have developed a gradient based segmentation method to identify the helices of a protein. Using the identified helices as geometrical constraints, we developed a method to analyze a large set of possible model structures generated by Rosetta, a structure prediction software. The possible sequence assignment and topology of the visualized helices can be derived. (There is an opening for Postdoctoral Research Position)