Multivariate Statistical Analysis of Protein Crystal Unit Cells

The unit cells of protein crystals can vary due to conformational changes of the protein. Electron crystallography can only give an average picture of the contents of a unit cell. However, it is the conformational changes that make the protein functional and they are therefore especially worth studying. Using multivariate statistical analysis it is possible to analyse the differences between different unit cells in the same crystal. The proposed project involves dissecting high resolution (but very noisy) electron microscopic images into individual unit cells (for example by cross correlation methods) and then classifying the resulting set of unit cell images by pattern recognition algorithms.