应用454测序技术分析菌群结构的方法学分析-methodological analysis of microbial community structure by 454 sequencing technology.docxVIP

II II 行的 454 分析方法，如 CVTree 和 BLAST-based 方法进行比较，结果 显示我们的分析流程获得的菌群结构和样本关系与 16S rRNA 全长的 结果更为相近。 关键词：菌群结构；454 高通量测序；454 测序的分析方法；重复性； 可靠性；CVTree 方法；BLAST-based 方法 PAGE PAGE IV The method study of the application of 454 pyrosequencing on microbial community analysis ABSTRACT The microbioal communities exist in all parts of the biosphere. Their functions are determinded by the composition. The new generation sequencing technology such as 454 pyrosequencing provide new way for understanding of the diversity and composition in microbial communities with the advantage of high-throughput ability, cost-efficiency and automated facility. However, how to efficiently use and to analyze these massive 454 sequencing data still remains a question. This study focuses on three issues. Initially, based on the real data from our lab, we simulated the 454 reads of V1-V2, V3 and V6 genes to obtain the composition of the microbioal community in those regions and compared them with the one obtained by 16S rRNA full length sequences, respectively. The result showed that all the three hypervariable region genes could capture the comparable composition information to the full length sequences while the V6 genes carry taxonomic instability at the level of genus. Then, based on the current software and methods, an integrated and improved data analysis pipeline including the quality control, alignment, distance matrix calculation, operation taxonomy unit definition, diversity estimation，taxonomy assignment and phylogenetic analysis procedures was established. The application of the pipeline on the real 454 reads showed there was no significant difference between 16S rRNA full length sequences and 454 reads. In addition, good reliability and repeatability of dominant taxa at each taxonomic level were observed. However low abundant taxa (5%) were detected unstable when the sequencing depth is about 800 reads per sample. Finally, compared with the CVTree and BLAST-based method, bette