Use of multivariate statistics in analysing the images of biological macromolecules

We have developed a new technique of analysis that allows automatic classification of molecule images according to subtle differences. Computer alignment and multivariate statistical methods were used to analyze electron micrographic images of horseshoe crab hemocyanin half-molecules. The molecule projections fell into four distinct classes related to four different positions of the molecule on the grid. Averages obtained for each images subset are interpreted in terms of a three-dimensional model arrangement for the four subunits forming the half-molecule.