Christine Lo CS 267 Assignment 0 Protein Folding I. Bio I am a fourth year undergraduate EECS major. Currently I am doing computational biology research under Professor Karp. I hope to gain from this class a better knowledge of the different types of parallel applications and also how to develop algorithms for parallel computing. II. Problem Background Proteins are a linear chain of amino acids that folds into a 3-D shape that ultimately determines the function of the protein. In the study of protein folding, there are two main aspects. The first is in predicting the protein shape and the second is determining the pathway of how the protein folds to assume its shape. Predicting the shape of a protein is directly related to understanding the function of the protein. Thus, it is very useful for the development of medicine. Understanding the folding pathway is also important for understanding misfolded proteins. These proteins are associated with many diseases such as Alzheimer’s, Huntington’s, and Parkinson’s. Experimentally determining the proteins shape involves the X-ray crystallography process which is expensive and difficult. Thus, scientists have begun to rely on computational methods for studying protein folding. Below is a picture of the simulation a protein from its partially unfolded state to its folded state. III. Parallel Application Protein Folding is a very active research subject and is an ongoing project on several major supercomputers. The University of Washington’s Rosetta@Home a Distributed Computing Cluster dedicated to work on protein folding prediction. It is focused on predicting theChristine Lo structure of proteins.The prediction of the protein structure doesn’t require as much computational power as determining folding pathways. Rosetta@Home has a throughput of around 89 teraFlops. Probably the most famous Distributing Computing Cluster is Stanford’s Folding@Home. Unlike Rosetta@Home, it is more focused on the folding process of a protein rather than predicting protein structure. People across the world can download and run their software to solve a different portion of the problem. All together Folding@Home currently has a throughput of around 4.7 petaFlops. The Blue Gene Watson (BGW) is the twenty-fifth fastest supercomputer on the Top500 list that also focuses on the determining the folding process. It is located at IBM’s Watson Research Center in New York. The following is a picture of the Blue Gene Watson. It has a computing power of over 91 teraFlops. 80% of the Blue Gene Watson’s computational capacity is dedicated to biological simulations, in which half of that is dedicated to protein folding. It runs an adaption of the software used for Rosetta@Home. The BGW has proven its success it both its accuracy in protein folding and its computing power. It has achieved a complete CASP7 prediction in less than three hours which is a significant improvement for a task that would normally take several weeks. CASP7 or Critical Assessment of Structure Prediction 7 (CASP7) is composed of researchers studying protein folding. It aims to achieve correctness in protein structure and folding pathway predictions by comparing computational results to experimental results.Christine Lo IV. References Wikipedia- Protein Folding http://en.wikipedia.org/wiki/Protein_folding Folding@Home http://folding.stanford.edu/English/Main Top 500- Blue Gene Watson http://www.top500.org/system/7466 Blue Gene http://www.research.ibm.com/journal/sj/402/allen.pdf Rosetta@Home http://boinc.bakerlab.org/rosetta/ San Diego Supercomputer Center
View Full Document
Unlocking...