Single-particle reconstruction (SPR) electron microscopy is a powerful technique for obtaining the 3D structure of macromolecular complexes in near-native states. A major problem in SPR is that flexibility gives rise to multiple conformations when the macromolecular particles are frozen in vitreous ice. Traditional SPR assumes that macromolecular particles are rigid, and gives good reconstructions only when all particles have the identical structure. We propose to develop new reconstruction methods which eliminate the rigidity assumption altogether. These methods can reconstruct a continuously flexible structure. Two classes of methods are proposed: (a) methods for exploratory analysis which detect the presence of structural conformational change and objectively determine the most dominant modes of flexibility, and (b) post- exploratory flexible particle reconstruction methods which reconstruct in detail the identified modes of flexibility. These methods are based on probability theory and preliminary results with actual data show that they offer significantly higher signal-to-noise ratio (SNR) reconstructions even when the data SNR is as low as -25db. Further, our methods provide a simple interpretation of particle flexibility. We will develop these algorithms and write software modules for incorporation in the SPARX cryo-EM processing environment. As a first application, we will use the software to process negative-staining and cryo-EM images of the human Dicer RNA-processing enzyme. As positive and negative controls, we will also reconstruct the structure of RNA Polymerase II and GroEL.

Public Health Relevance

Electron microscopy can provide detailed pictures of large molecular complexes from cells. Unfortunately, when the complexes are flexible, they take on different shapes at the instant they are frozen for imaging;in this case the reconstructions are blurred, because they are formed as averages from images of many individual complexes. We propose to develop a new approach for mathematically modeling the flexibility and allowing high-quality reconstructions to be made from flexible molecules such as human Dicer.

National Institute of Health (NIH)
National Institute of General Medical Sciences (NIGMS)
Research Project (R01)
Project #
Application #
Study Section
Special Emphasis Panel (ZRG1-IMST-L (90))
Program Officer
Flicker, Paula F
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Yale University
Schools of Medicine
New Haven
United States
Zip Code
Huang, Chenxi; Tagare, Hemant D (2016) Robust w-Estimators for Cryo-EM Class Means. IEEE Trans Image Process 25:893-906
Tagare, Hemant D; Kucukelbir, Alp; Sigworth, Fred J et al. (2015) Directly reconstructing principal components of heterogeneous particles from cryo-EM images. J Struct Biol 191:245-62
Kucukelbir, Alp; Sigworth, Fred J; Tagare, Hemant D (2014) Quantifying the local resolution of cryo-EM density maps. Nat Methods 11:63-5
Huang, Chenxi; Tagare, Hemant D (2014) Robust estimation for class averaging in cryo-EM Single Particle Reconstruction. Conf Proc IEEE Eng Med Biol Soc 2014:3329-32
Shigematsu, H; Sigworth, F J (2013) Noise models and cryo-EM drift correction with a direct-electron camera. Ultramicroscopy 131:61-9
Kucukelbir, Alp; Sigworth, Fred J; Tagare, Hemant D (2012) A Bayesian adaptive basis algorithm for single particle reconstruction. J Struct Biol 179:56-67