The conventional negative-stain preparation method for electron microscopy, in which biological macromolecules are contrasted using heavy metal salts (such as uranyl-acetate), is a simple and fast technique which has helped visualize hundreds of different molecular structures. Computer analysis of such negatively stained images of individual (i.e., non-crystalline) macromolecules using statistical pattern-recognition techniques has revealed considerable new structural information. Negative staining, however, has some disadvantages: the specimens are often severely flattened (as much as 25%-75%), they often exhibit strong preferential attachment of the molecules to the supporting carbon foil, and the molecular images may be difficult to interpret due to the relatively complex nature of the interaction between molecules and stain. Embedding biological macromolecules in a layer of vitreous ice (actually: "vitreous water") represents an attractive alternative preparation method which mimics the natural environment of these molecules. The processing of ice-images often requires special computational approaches such as: multivariate statistical classification of aligned images or of "invariant functions" derived from the unaligned images; alignment of images belonging to a specific class of images, determination of the spatial orientations of the projection images relative to each other ("angular reconstitution"). In this paper, we discuss our own overall single-particle structure analysis approach and highlight some new methodological developments in this context.
van Heel, M.; Winkler, H.; Orlova, E.; and Schatz, M.
"Structure Analysis of Ice-Embedded Single Particles,"
Scanning Microscopy: Vol. 1992
, Article 3.
Available at: https://digitalcommons.usu.edu/microscopy/vol1992/iss6/3