Resource for the Visualization of Biological Complexity (RVBC)
Classification of heterogeneous data in single-particle reconstruction
Single-Particle Reconstruction is the reconstruction of macromolecular complexes from thousands of low-dose images that show the molecule in different views. The tacit assumption, that the molecule exists in the specimen in the same conformation or binding state, is often not fulfilled. In those cases, the reconstruction is blurred and fails to reflect the true structure of any of the conformations. An example for such a complication is the ribosome, which upon binding with EF-G (stabilized with an antibiotic) changes its conformation. If only 50% of the molecules are bound with EF-G, then the reconstruction of the complex is adversely affected in two ways: (1) the density of the ribosome is blurred by the averaging of data originating from the two conformations, and (2) the density relating to the bound EF-G is only half the value found for ribosomal proteins, and will therefore be invisible at normal density thresholds.
This core project has as its aim the systematic exploration of methods to sort out heterogeneous molecule populations, so that subpopulations are found that can be separately reconstructed. To this end, collaborations have been started with experts in the field: (1) Jose-Maria Carazo, at the Centro Nacionale de Biologia in Madrid; and (2) Gabor Herman, at CUNY in New York City.
Both simulated and experimental cryo-EM data are being used to test the efficacy of several approaches to classification. In addition to in-house data on ribosome complexes and ryanodine receptors in different binding and switching states, we have obtained data from other labs that pursue cryo-EM of macromolecular complexes. Although both supervised and unsupervised classification have be tried, the emphasis of the development will be on novel methods of unsupervised classification.
This core project is closely related to the core project on time-resolved cryo-EM of macromolecular complexes, which is giving various mixtures of complexes at different time points of a reaction.
References
Scheres SH, Gao H, Valle M, Herman GT, Eggermont PP, Frank J, Carazo JM. (2007) Disentangling conformational states of macromolecules in 3D-EM through likelihood optimization. Nature Methods 4:27-9.
Fu J, Gao H, Frank J. (2007) Unsupervised classification of single particles by cluster tracking in multi-dimensional space. J Struct. Biol. 157:226-39.
Penczek PA, Frank J, Spahn CM.(2006) A method of focused classification, based on the bootstrap 3D variance analysis, and its application to EF-G-dependent translocation. J Struct. Biol. 154:184-94.
Herman GT, Kalinowski M.(2008) Classification of heterogeneous electron microscopic projections into homogeneous subsets. Ultramicroscopy 108:327-38.