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

16 CHAPTER 1. INTRODUCTION
Our application programming framework libshore was implemented as a common foundation
to all SHORE processing modules. In addition to elementary functionality such as automatic
input and output decoding and encoding, algorithms for indexing sequencing-related data and
alignment of biological sequences are provided. Furthermore, we developed a generic processing
framework with the aim of facilitating breakdown and implementation of rather complex
sequencing data analysis algorithms as collections of manageable and self-contained processing
modules. We further extended on this framework for straightforward parallelization of basic
sequencing data processing.
The fundamental data analysis workow introduced with the initial version of SHORE consists
of subsequent application of read ltering, read mapping and application-specic analysis
modules. While this basic approach has been maintained, it has been extended in many areas
for simplication of a variety of routine tasks. An overview of the modied workow is therefore
given in section 2.1.
Signicant modications to core pipeline modules concern raw sequencing read import and
ltering as well as read mapping. Data import was modied to enable recovery of raw data
or iterative data set re-ltering (section 2.3), and complemented with extended functionality to
address e. g. the more and more widespread use of a variety of library multiplexing and further
adapter ligation protocols. For this purpose, exible sequence-specic read partitioning and
clipping facilities have been added (sections 2.4, 2.5). SHORE's read mapping module initially
constituted a basic parallelization wrapper providing a unied interface to dierent short read
alignment tools. To take advantage of an increasing diversity of read mapping algorithms, features
have been added to enable integration of the results obtained using dierent alignment
algorithms or parameters (section 2.6).
Application-specic data analysis modules initially implemented in SHORE mainly addressed
the assessment of genomic variation. This functionality has largely been maintained, with only
slight modications to take advantage of the libshore infrastructure. A focus of this work however
was on quantitative application of high-throughput sequencing, and primarily ChIP-Seq
data analysis. Within the SHORE analysis environment, we have implemented an enrichment
detection module targeted at the analysis of transcription factor immunoprecipitation data (section
2.7). In comparison to other approaches, our method emphasizes robustness towards read
mapping artifacts and simplies handling of replicate experiments. Our ChIP-Seq enrichment
detection module is complemented by congurable auxiliary utilities allowing composition of custom
workows applicable to various expression proling or enrichment detection applications.
In conclusion, with this work the SHORE software has been developed from an analysis pipeline
consisting of a set of largely independent submodules targeted primarily at genomic variation
assessment into a tightly integrated, general-purpose sequencing data analysis suite supporting
a set of coordinated, interdependent features. The SHORE sequencing data analysis suite is open
source software freely available 1 under the GNU General Public License (GPL) version 3.
1 http://shore.sf.net