A GPU-based high performance system for discovering consensus domain architecture and functional annotation of protein families

点击翻译按钮获取中文摘要

The list of organisms with completed genome sequence is continuously growing and this has led to the identification of thousands of genes whose function is still unknown. These genes could potentially be involved in important biological cell functions and could represent important targets for diagnostic and pharmacogenomics studies and be of industrial and agronomical importance. A major undertaking for biology is therefore that of identifying the function of these uncharacterized genes on a genomic scale. The challenge for bioinformatics is then to develop algorithms that, given a gene, can predict a hypothesis for its function.Comparisons of sequences from complete genomes have revealed that gene duplication, divergence and rearrangement are predominant mechanisms that drive the expansion of the set of proteins of a given organism during evolution. This means that proteins can be grouped into families, where members are likely to perform similar functions. The identification of these protein families is therefore central as it can provide important clues for the function of proteins.Proteins are often composed of several domains. A domain is segment of protein sequence that can evolve independently of the rest of the protein chain. Each domain forms a compact three-dimensional structure and it can appear in a variety of different proteins. Protein function depends on the mutual interplay between the distinct domains and the links between them. In other words, protein function depends on the domain architecture of the protein.Therefore we would like to have a tool that can group proteins into families according to their architecture: all proteins with the same architecture should belong to the same group. The development of such a tool is exactly the goal of this project. Moreover the tool that we plan here will also be able to suggest possible functional roles for the various architectures.Our tool is aimed at working on very large sets of proteins. The amount of calculations for problems of this size is only feasible by taking advantage of the latest advances in graphical processing unit (GPU) technology. Modern GPUs are very efficient for graphics but their highly parallel structure makes them extremely effective for algorithms where processing of large blocks of data is done in parallel - even more effective than general-purpose CPUs. The use of GPU technology will allow us to create a web application that will be used by scientists to obtain the architectures for very large set of proteins together with possible functional roles for the various architectures. Importantly, we shall periodically run our system on the major genomes available and we will thus be able to through our web server architectures and relative annotation for all the proteins in those genomes. All these web services will be made freely available to the scientific community.