Develop Pattern Match Analysis Platform for Large-Scale Metagenomic and Transcriptome

Next-generation sequencing technologies have been an important research filed. With the development of sequencing technology, metagenomics and transcriptome also become popular research topics. For metagenomics, we can obtain reads of all sample data in any environment by using sequencing technology at first. Then, we can observe biological diversity under microscopic. Finally, we can understand the microbial world. After sequencing a large number of biological genomes, the next work is how to annotate the function of these genes from these genomes. By gene microarray and analysis techniques, transcriptome can do a comprehensive study of gene expression in the genome. Therefore, the goal of this integrated project is to construct an analysis platform for metagenomics and transcriptome. The main works of integrated project are to develop basic techniques and apply this platform for real applications. The goal of this sub-project is to develop a pattern match analysis platform for large-scale metagenomic and transcriptome. The main works of this sub-project are to construct a pattern match analysis platform under multi-core CPU and accelerators (GPU and Intel XEON Phi), develop a series of blast tools (ex. blastp and blastn) in this platform, build a big data management services, and design a visual analysis platform. First year, we will construct a pattern match analysis platform under multi-core CPU. We also will construct a pattern match analysis platform under multi-GPUs. A series of blast tools based on GPU environment will be developed, and a big data management services will be built in the platform. Second year, we will construct a pattern match analysis platform under Intel XEON Phi, and Phi-based blast tools will be developed and compared with GPU-based version. The performance of a big data management services will be evaluated and improved, and we also will design a visual analysis platform. Third year, we will construct a pattern match analysis platform under GPU collaborated with Phi environment. A series of blast tools will be developed by integrating GPU and Phi computing capabilities. We will propose a technology to detect long sequence repeats in the genome. A integrated platform for big data management and visualization services will be built.

Project ID:PB10308-3330
External Project ID:MOST103-2221-E182-027