Faculty Interaction

Wen Jiang

 

Department of Biological Sciences
Purdue University

Brief Project Description:

Anybody who's had a cold knows the problem with viruses, but the problem goes well beyond a clogged nose and bad cough. Viruses also cause diseases—from AIDS and Ebola to SARS and Swine Flu—deadly for millions of people around the world. Viral infections remain one of the major threats to human health.

Many viruses consist of a highly symmetric icosahedral protein shell that encapsulates the genetic materials. A high-resolution, three-dimensional schematic of that structure, if available, could provide important insights for understanding viral infections and developing more effective prevention and treatment strategies.

Wen Jiang, using ITaP research computing resources, is generating such images, along with knowledge about the structure of viruses that could eventually lead to new ways of battling them by targeting their basic functionality.

Jiang, a member of Purdue's structural biology group, uses electron cryo-microscopy, or cryo-EM, to look at the inner workings of large molecular machines such as viruses, the target system on which his lab mostly focuses for now.

The goal is to solve the 3-D structure of a virus at near-atomic resolution, the better to understand its mechanisms, including how a virus infects a host. Jiang's lab has reached 3.65 angstroms with the epsilon15 bacteriophage, a model virus that infects bacteria. A human hair is about a million angstroms in diameter. The researchers hope to get to 3 angstroms in coming years

The technique also can offer a window into the structure of the molecular machines permeating our bodies, such as DNA molecules and the myriad proteins that fulfill key biological roles. Beyond biology, it could be employed in working with a variety of nano structures.

The methodology being developed will benefit others using cryo-EM for whatever purpose. Jiang also is making the key image processing algorithms developed by his group available for free to academic researchers.

A standard technique, X-ray crystallography, is difficult to apply to large or complex structures like a virus because they're tough to crystallize effectively. Electron cryo-microscopy bypasses the crystallization step. It uses electrons to bombard virus samples purified in a solution very similar to the environment in a host cell and super cooled to liquid nitrogen temperatures, about 180 degrees below zero. That doesn't crystallize a sample, but does make it as if it were “frozen in glass,” as Jiang puts it, yet still alive and infectious.

The signals gleaned from electrons interacting with a sample are the raw material of the structural images Jiang and colleagues create. Essentially, the researchers use image processing to assemble a host of 2-D images captured from different points of view into 3-D images. The method is akin to a computerized axial tomography, or CAT, scan or the stereo viewpoint human eyes use to render the world in three dimensions.

The researchers image a virus in individual “particles.” The particles, as many as 100,000, are imaged as 2-D views then aligned to create a 3-D image of the structure in a process covering all possible points of view. Meanwhile, each particle at high resolution may weigh in at 1,000 by 1,000 pixels. It's a staggering amount of data.

To get the computing power required to process that data, Jiang's lab draws cycles from Purdue's Condor pool. The distributed computing system operated by ITaP's Rosen Center includes tens of thousands of machines connected to the Purdue network on and off campus and made available for research when idle, both at Purdue and on the National Science Foundation's nationwide TeraGrid network.

The work is computationally intensive, but the entire task can be broken into many independent small jobs, perfect for parceling out to the thousands and thousands of processors available in the Condor pool. Reaching even higher resolutions will require significantly more computing power. The move from 4.5 to 3.65 angstroms nearly doubled the amount of data.

Jiang works with colleagues in computer science and other fields to improve his technique and the Condor technology, for example, through checkpointing, which could allow a job to be interrupted and shifted from machine to machine seamlessly, without losing work already done. Improved methods for distributed storage also are of interest given the large data sets the cryo-EM reconstruction process generates.

 

More information:

http://jiang.bio.purdue.edu

Last Updated 6/09/09