World Wide, Web-Enabled Experiments in Protein Structure Prediction
University of Maryland Biotechnology Institute
About the Lecture
At the molecular level, most biological function is expressed through the complex three dimensional structures of protein molecules. These structures are determined by the information in the linear amino acid sequence alone, information that is in turn derived from the sequence of bases in the DNA of a gene. This transformation from a linear information string to a complex three dimensional structure capable of biological function lies at the heart of all living processes. A major computational challenge is the development of methods that can calculate the spatial positions of the atoms of a folded protein, given only the amino acid sequence.
For the past thirty years, many research groups have worked on solving this problem, using the conventional organization of contemporary science, with small groups working largely in isolation, and publishing in peer reviewed journals. Although much excellent work has been done, progress has been gradual, with many false leads receiving acceptance.
The advent of universal electronic communications, particularly the web, has made it possible to consider different paradigms for the conduct of scientific research. For protein structure prediction, we have explored the use of a model in which many different research groups agree to work on predicting the structure of the same set of proteins, followed by an extensive comparison of the quality of the results obtained from the different methods. The outcome of these experiments will be described, both in terms of what we have learned about predicting protein structure, and what we have learned about the advantages and disadvantages of this way of doing research.
About the Speaker
John Moult is a professor at the University of Maryland Biotechnology Institute and fellow of the Center for Advanced Research in Biotechnology. He received a Doctor of Philosophy from the University of Oxford in 1970. He has published frequently on the subject of protein folding and related areas, and has worked to apply mathematical models to problems of chemical analysis and prediction.
President Garavelli called the 2101st meeting to order at 8:22 p.m. on February 19, 1999. The Recording Secretary read the minutes of the 2100th meeting and they were approved.
The speaker for the 2101st meeting was John Moult of the Center for Advanced Research in Biotechnology, University of Maryland Biotechnology Institute. The title of his presentation was, “World Wide Web-Enabled Experiments in Protein Structure Prediction.”
Mr. Moult presentation covered two related issues. The first issue was specific and addressed the “protein structure problem.” The second issue was more general and addressed “how we do science in the post cold war world” or stated another way, “how we approached the protein structure problem in the post cold war world - with the world wide web available to us.”
Considering the protein structure problem, hundreds of thousand of different proteins comprise the biochemical molecules that make up cells, organs and organisms. The various proteins are comprised of characteristic strings of different amino acids. As we all know, the larger structures of cells, organs and organisms put themselves together to form unique structures. Similarly, proteins put themselves together uniquely. This protein “putting together” process is called “folding.” The resultant unique structures produced by folding are determined by the information contained only in the linear sequence of the protein's amino acids. Just how the proteins do this assembly work is the “protein folding problem” and it is one of the most important unanswered questions in the life sciences.
For the past thirty years, multiple research groups have struggled with the conventional tools of contemporary science. This approach essentially mandated small groups working largely in isolation and publishing in peer reviewed journals. Although much excellent work was done using this approach, progress was gradual, with many false leads receiving premature acceptance.
However, with the recent advent of universal electronic communications, particularly the web, it became possible to consider different paradigms for doing scientific research. This is Mr. Moult's second issue, doing scientific research in the post cold war world. The specific research considered was the quest for effective tools for protein structure prediction. The research was even given its own name, the “Critical Assessment of methods for protein Structure Prediction” with the acronym, “CASP.” The CASP experiment approach has multiple parts. The problem starts out with many protein predictions targets, multiple predictions from each player, many players taking part and authoritative evaluation. This novel approach continues with collecting information about the to-be-solved protein structures from the experiment community, passing the information on to the prediction community, collecting the structure models, evaluating the models, meeting and discussing the results and finally — releasing the results.
Actual implementation of this approach began in 1994 with many different research groups agreeing to work on predicting the structure of the same set of proteins. This was followed by an extensive comparison of the quality of the results obtained from the different structure prediction methods. There are three principal prediction methods: the traditional comparative modeling method which is useful but has problems and will likely continue to have problems, the threading method which holds much promise and the ab inuito method which also holds promise. All three methods provided some approximation of structure. The comparative method and the threading methods also provided some understanding of function and stability. However, none of the methods as yet can provide the needed accuracy needed for drug design. This initial CASP endeavor was referred to as CASP 1, as we might have expected since it was followed by the expanded endeavors of CASP 2 and CASP 3.
When pressed to predict the future of our ability to predict protein structures, Mr. Moult projected that within two years, we will likely know everything that goes on in simple cells, but we will not understand it. This remains as another piece of the protein puzzle.
Mr. Moult then closed his presentation and kindly answered questions from the floor. President Garavelli thereupon thanked Mr. Silver for the society, announced the next meeting and made the usual parking announcement. He then adjourned the 2101st meeting to the Social Hour at 9:27 p.m.