A team from the MU Department of Computer Science received a $1.4 million grant from the National Institutes of Health this summer to help fund protein structure prediction research.
Professors Dong Xu and Yi Shang, along with their research group, have worked on predicting protein structure since 2004. This is the third time they have received a grant from NIH. Last year they received $945,300.
Proteins serve many vital functions in the human body, such as speeding up chemical reactions, communicating between cells and their environment, movement, defense, energy storage, structure and transportation. They fold up in unique structures, and the structures determine the protein’s function. If scientists can map out every structure of all the proteins the body makes, it would provide them with important information for things such as battling diseases and designing new drugs.
It’s a process that Xu describes as “highly challenging.” He said scientists can’t rely on current computational methods to predict consistently accurate structures. New forms of software tools that predict protein sequences based on next-generation sequencing seem to generate precise structural prediction.

According to Xu, the project answers the need for new tools by developing new methods into providing precise structures. The research group is putting together its own unique tool.

“We comprehensively integrate bioinformatics techniques, graph and network, theories, computational algorithms, global optimization methods and statistics evaluations to develop new algorithms and improve our protein structure prediction tool, which is uniquely our own and will hereafter be referred to as MUFOLD,” Xu said in an email.

Xu said MUFOLD takes a new approach to predicting protein structure, converting the “prediction problem” to a “graph-realization problem” and using a “global optimization approach of multi-dimensional scaling” in order to solve it. This contemporary method provides scientists a more efficient and deeper picture of protein structure. The goal of MUFOLD, once fully developed, allows experimental biologists to understand protein structure and function and make hypotheses.

Both Xu and Shang earned their Ph.D.s at the University of Illinois in the 1990s. At the same time, they began talking about protein structure prediction. Shang, who worked at the Xerox’s Palo Alto Research Center in 2002, examined multidimensional scaling while he was working on wireless sensor networks. The scaling involved data analysis techniques using multiple algorithms and displaying the data geometrically. This technique is now being implemented with MUFOLD.

In 2008, the research team first participated in Critical Assessment of Protein Structure Prediction. CASP is an international competition held biennially. During the 2010 CASP, MUFOLD scored high in the human/server prediction and the quality assessment categories.

Along with Xu and Shang, the research team consists of MU physics professor Ioan Kosztin, research scientist Jingfen Zhang, graduate student Zhiquan He and graduate student research assistants.

“We are confident that we will deliver a highly useful tool for the
community,” Xu said. “We have made a lot of progress towards this goal.”