“Signature: A Particle Data Analysis System for Macromolecule Electron Microscopy”
James Zhi Chen – Brandeis University
Signature is data processing software for macromolecule electron microscopy (EM). The system has two major components: a particle selection module and a particle analysis module. The graphical user interface of the program provides informative data display to facilitate and expedite EM data processing.
The particle selection module supports both automatic and manual operations. High-throughput selection can be achieved by automated screening, which applies a hierarchical template-matching algorithm to identify the molecules of interest in electron micrographs. The output is reproducible and is not susceptible to subjective bias. The algorithmic screening can operate in a batch-processing mode that can greatly enhance productivity through scripting and distributed computing. Particles can also be interactively selected through the graphical interface of the program.
The data analysis module provides functions for particle alignment, particle classification, 2D image processing and 3D density map display/editing. A few newly developed analytical tools, such as digital gel filtration, have also been incorporated into the system.
Signature has been validated with both synthetic and experimental EM datasets. A beta-release of the program is currently available for Linux, Mac OS and MS-Windows computing platforms, and it is being tested at several EM research groups. The program web page is at www.brandeis.edu/~jzchen/Signature.
In this talk, I plan to introduce and focus on the particle selection function of Signature. I will discuss the algorithm as implemented and demonstrate a few applications.