Toxicogenomics – A Drug Development Perspective - Genomic Biomarkers for Pharmaceutical Development - page 140

the form of bulk downloads of array data from each experiment type: in vivo single dose liver
and kidney (E-MTAB-799), in vivo repeat dosing liver and kidney (E-MTAB-800), in vitro rat
hepatocytes (E-MTAB-797), and in vitro human hepatocytes (E-MTAB-798). This citation also
provides a description of the Japanese Toxicogenomics Project.
Meanwhile, one of the integrated EU Framework 6 Projects entitled Predictive Toxicology
(PreTox) tackled toxicogenomics using a systems toxicology approach. The project was coor-
dinated by the European Federation of Pharmaceutical Industries and Associations (EFPIA),
a body representing the research-based pharmaceutical industries and biotech Small and
medium-sized enterprises (SMEs operating in Europe. It is partly funded by the European
Commission Life Sciences, Genomics and Biotechnology for Health Priority with a grant of
€8M aimed at assessing the value of combining results from 'omics' technologies (genomics,
proteomics and metabolomics) together with the results from conventional toxicology meth-
ods, for more informed decision making in pre-clinical safety evaluation. The effects of 16
test compounds were characterized using conventional toxicological parameters and 'omics'
technologies. The three major observed toxicities, liver hypertrophy, bile duct necrosis and /
or cholestasis, and kidney proximal tubular damage, were analyzed in detail [225,226] . The
combined approach of 'omics' and conventional toxicology has proved to be a useful tool for
mechanistic investigations and the identification of putative biomarkers [29] .
6.7 BI OINFORMATICS CHALLENGES IN TOXICOGEN OMICS
The exponential growth in the amount of toxicogenomics data being generated poses
both challenges and opportunities, i.e., data management and extraction of useful infor-
mation from these data, and the development of tools and methods capable of transform-
ing these data into knowledge. Both the generation and validation of hypotheses require
not only a comprehensive description of the different components of an experiment (such
as cell / organ exposure, sample collection, and processing components of the experi-
ments) in the database, but also complex computational and bioinformatics approaches.
Bioinformatics is an interdisciplinary field that develops and improves upon methods for
storing, retrieving, organizing, and analyzing biological data [227] . It is a bridge between
observation (experimental data) in diverse disciplines of biology and genomes and extrapo-
lation of information, by computational means, about how the systems and processes func-
tion [228] . One objective of bioinformatics analysis is to extract useful knowledge from the
flood of data, including biological texts, for the purpose of further analysis leading ulti-
mately to useable knowledge [36] and [229-231] .
Due to the nature and characteristics of the diverse techniques that are applied for biologi-
cal data acquisition, and depending on the specificity of the domain, biological data might
require a number of preparatory steps prior to analysis. These are usually related to the
selection and cleaning, preprocessing, and transformation of the original data. The data pre-
processing task is subdivided into a set of relevant steps that could improve the quality and
success when applying, for example, machine learning techniques [232,233] . They reine / dep-
urate the data to make certain data mining operations more tractable. There are three well-
known data preprocessing topics that are among the most applied. These are missing value
imputation, data normalization, and discretization [234,235] .