An Integrated Data Analysis Suite and Programming ... - TOBIAS-lib

1 Introduction
Next-Generation Sequencing has evolved into a powerful tool for many areas of biological
science, but has also introduced many new challenges related to data analysis
and computing infrastructure into the eld. Following a brief recapitulation of highthroughput
DNA sequencing technology, this chapter highlights important areas of
application while discussing previous and related work. We outline general sequencing
data analysis methodology and conclude by establishing the motivation for this
work.
1.1 High-Throughput DNA Sequencing
Widely deployed for less than a decade, the impact of parallel next-generation sequencing technology
on genetic research has already been considerable. The novel high-throughput sequencing
approaches are characterized by massive parallelism implemented in a single device. From this
design results the key advantage of a signicantly reduced cost per sequenced base, which has
allowed to quickly displace the previously predominant Sanger sequencing method in many areas
of application.
Sanger sequencing, the prevalent sequencing method for the majority of time since its introduction
in 1977 [2], needed to rely on extensive automation in dedicated sequencing centers to
achieve time and cost eective readout of large amounts of sequence. Most prominently, this was
put into eect during the eort of generating the rst draft sequence of the human genome [36].
By contrast, high-throughput sequencing instruments are designed to analyze thousands to millions
of DNA molecules simultaneously, and thus enable even smaller institutions to produce
large quantities of sequence data on site.
Secondary to elucidation of DNA primary structure, sequencing utilized as a random sampling
device delivers quantitative clues on sample composition. In this capacity, it has been
used for diverse purposes such as inference of DNA methylation levels [7] or the analysis of environmental
samples [8]. In concert with the economical advantage over the dideoxynucleotide
chain-termination method, parallel DNA sequencing becoming widely accessible to researchers
has served to promote such deep sequencing approaches that open up whole new areas of application
beyond the domain previously occupied by Sanger sequencing.
As an alternative to microarray technologies, for example in gene expression proling or
chromatin immunoprecipitation assays, deep sequencing overcomes fundamental technological
restrictions such as probe resolution and probe saturation. Furthermore, without the inherent
requirement of a-priori knowledge of sequences to be detected or quantied, development of novel
protocols of application has been furthered to transform high-throughput sequencing methods
into a versatile tool for research.
1