Automated sequence preprocessing in a large-scale sequencing environment

Autor: LaDeana W. Hillier, Michael C. Wendl, Dave Hodgson, Simon Dear
Rok vydání: 1998
Předmět:
Zdroj: Genome research. 8(9)
ISSN: 1088-9051
Popis: Increased automation in the Human Genome Project (Watson 1990) continues to be critical in meeting projected goals and minimizing costs. Whereas technology development and hardware improvements are responsible for many gains, informatics issues also remain critically important (Marshall and Pennisi 1996). One area in which this is especially true is processing shotgun data, that is, converting raw fluorescent-gel images into assembled sequence. Managing the voluminous amount of data is difficult and requires appropriate computer software to maintain the systematic flow, organization, and quality-control necessary for successful large-scale shotgun sequencing (Wilson et al. 1994; Mardis and Wilson 1997). Most proprietary systems have proven to be less than adequate, however, a number of excellent Unix software tools have been developed to aid in automating various steps. For example, programs are now available that perform lane retracking calculations on multilane gel images with only a minimum of human intervention (Cooper et al. 1996). Moreover, a number of robust assembly algorithms and editors have been implemented and presently are in widespread use, for example, gap (Bonfield et al. 1995), phrap (P. Green, pers. comm.), consed (Gordon et al. 1998), and fakII (Larson et al. 1996; Myers 1996). The time, effort, and cost saved by such software is significant. Sequence preprocessing, also called preassembly (Bonfield and Staden 1995), is the transformation of raw trace signals to assembly-ready sequence and is flanked by lane retracking and assembly (Fig. ​(Fig.1).1). It includes tasks such as converting the raw trace file from proprietary to standard form (Dear and Staden 1992), deriving template information, base-calling, vector screening, quality evaluation and control, disk management, and associated tracking and reporting operations (Fig. ​(Fig.2).2). Not only does it represent a significant computational effort, but sequence preprocessing is also challenging in the sense that it must handle many clones (also called sequencing projects or simply projects) being sequenced simultaneously. The importance of analyzing preprocessing results and generating informative reports also cannot be underestimated.These reports are usually the only way to quantitatively evaluate incremental modifications of sequencing protocols that are often required. For high-throughput labs, an organized, automated sequence preprocessing strategy is therefore a necessity. Figure 1 Overview of large-scale data processing. Data are gathered from ABI sequencing machines (models 373 and 377) on a Macintosh and ported to the Unix network. All subsequent operations occur strictly in the Unix domain including lane retracking, sequence ... Figure 2 Overview of serial steps in sequence preprocessing. Each step represents an independent component module, some of which are simply wrappers for system programs. Modules can be added, deleted, or modified as needed. Template information refers to auxiliary ... Because of its fundamental importance in large-scale sequencing, the preprocessing problem has given rise to a number of Unix-based software systems using a wide range of paradigms. For example, pregap (Bonfield and Staden 1995) is a Bourne shell wrapper script designed around the Staden Package (Staden 1996) and the gap assembler (Bonfield et al. 1995). It is a component-based approach using the experiment file format (Bonfield and Staden 1995) for storing read-based information. The script allows some customization on the part of the user, but does not provide extensive capability for input validation and graphic input and output analysis. GRM (Lawrence et al. 1994) is an integrated graphic system based upon the object-oriented methodology. It is implemented in a number of languages including “C”, Smalltalk 80, and VisualWorks. The system allows interaction only through interfaces and supports various sequence assembly programs. Another program, hopper (Smith et al. 1997), is a Perl implementation designed around the University of Washington genome tools (P. Green, pers. comm.). Like pregap, it is component-based so individual pieces can be replaced with new or different tools. It also performs some input validation and automatically generates reports. The diversity of available preprocessing programs partially reflects the varying needs of individual sequencing laboratories and centers. Practices, protocols, lab organization, and file structure all have a bearing on designing a sequence preprocessing strategy. At the Genome Sequencing Center (GSC), we, along with collaborators from the Sanger Centre, began developing our current strategy and Unix-based Perl implementation 3 years ago. It has since matured into a complete system called the Genome Automated Sequence Preprocessor (GASP). Designed around our needs for large-scale throughput, it manages the current production rate of ∼50,000 genomic reads per week at each laboratory.
Databáze: OpenAIRE
Externí odkaz: