Within the past five years, many resolution breakthroughs through single particle cryo-electron microscopy (cryo-EM) analysis has transformed it into a mainstream technology for high-resolution protein structure determination. In contrast to X-ray crystallography, cryo-EM avoids crystallization in preparing particle samples and thus brings the hope to solve many important macro-protein structures. However, single particle analysis, which loses each particle’s angular labeling by its nature, has brought the 3D structure determination to an even more challenging computation task compared to crystallography. In the past years, we have developed many novel algorithms to face these challenges and integrated them on a processing platform called ASCEP.
We will give brief introductions of these algorithms, including
(1) a dimension reduction method for cryo-EM images,
(2) a visualization tool for checking classification performance,
(3) a 2-D classification algorithm package.
Real data analysis on cryo-EM image sets will be presented.