UTILIZATION OF SPECIALIZED HARDWARE FOR DNA SEQUENCE ANALYSIS

DCRT购买了ABI Inherit系统，目前正在提供 作为NIH内部科学家的共享资源。科学家购买 来自ABI的Macintosh客户端软件，并通过 美国国立卫生研究院网络。Inherit大量使用专门的硬件，即Fast Data Finder(FDF)，一种能够扫描和比较的并行处理器 每秒超过1500万个字符。本系统的主要设计是 用于：(1)大中型测序项目的组装；(2)搜索 同源基因序列的基因和蛋白质数据库； 快速搜索基因基序，如调控元件。一个 结合在系统中的模式搜索语言允许非常 复杂的查询公式。DCRT正在寻求与ABI进行讨论 关于一项合作协议，允许DCRT访问端口的源代码 服务器软件安装到其他的Unix平台，包括NIH Convex。 关于ABI继承评估，关键的、定量的 对几种不同的商业计算机程序进行了比较 设计用于序列组装和分析。这项研究的目的 有两个方面：(1)在使用几个方面获得经验和专业知识 不同的序列组装程序；(2)对这些程序进行评估 关于它们在组装中的速度和准确性以及它们的易用性。六 已经检查了序列组装包。以评估速度和 序列组装的质量，大鼠多药耐药基因(RATMDRM， 5,254个碱基对)序列随机分裂成58个重叠 片段，每个片段的长度为200-400个基准点。从0%到15%的错误是 根据误差分布随机添加到这些片段中 在用来寻找装配的原始碎片中找到的。

An ABI inherit system was purchased by DCRT and is being made available as a shared resource to NIH intramural scientists. Scientists purchase Macintosh client software from ABI and access the Inherit system over the NIH network. Inherit makes heavy use of specialized hardware, the Fast Data Finder (FDF), a parallel processor capable of scanning and comparing over 15 million characters per second. This system is primarily designed for: (1) assembly of medium to large sequencing projects; (2) searching the gene and protein databases for homologous gene sequences; and (3) searching quickly for genetic motifs such as regulatory elements. A pattern search language incorporated in the system allows for very complicated query formulations. DCRT is pursuing discussions with ABI on a cooperative agreement to allow DCRT access to source code to port Server software to additional UNIX platforms, including the NIH Convex. In connection with the ABI Inherit evaluation, a critical, quantitative comparison was done of several different commercial computer programs designed for sequence assembly and analysis. The purpose of this study was two fold: (1) to gain experience and expertise in the use of several different sequence assembly programs; and (2) to evaluate these programs as to their speed and accuracy in assembly and in their ease of use. Six sequence assembly packages have been examined. To evaluate the speed and quality of sequence assembly, the rat multidrug resistant gene (RATMDRM, 5,254 base pairs) sequence was randomly split into 58 overlapping fragments, each 200-400 basepairs in length. From 0 to 15% error was randomly added to these fragments based on the distribution of error found in the original fragments that were used to find the assembly.