368
V. Mittard-Runte et al.
methods and graphical displays. Within the BioConductor packages there are normalization methods for two-colour arrays as well as for Affymetrix
R
arrays. As a
downside of this flexibility, the system requires a high level of expertise. User interaction is mediated by a command line interface, and complex actions might even
involve programming. Anyway, we would recommend gaining an insight into and
to trying to work with R and BioConductor, as this combination provides maximal
flexibility in the choice of methods and their combination and full control over the
analysis process. Also, novel methods are often first implemented as BioConductor
packages. We refer the reader to Gentleman (Gentleman et al. 2005) for further
reference.
9.4.3 Data Sharing and Public Repositories
Gene expression experiments involve highly complex protocols and produce highvolume data. Compared with a genome sequencing approach the experimental
annotations need to be much more precise, as a good annotation of a functional
genomics experiment also involves environmental conditions. During growth, harvesting protocols, and further laboratory procedures, samples are transformed many
times. Also complex array designs are involved which can carry up to many
hundreds of thousands of features.
The experimental results of a microarray experiment need to be accompanied
with all this meta-information. Otherwise, it is almost impossible to interpret the
resulting data or even reproduce the experiment independently, which is still a difficult and underestimated task. To ensure scientific standards, almost all journals
in functional genomics now require the submission of microarray data to a public
repository prior to publication of results (Ball et al. 2004).
The most frequently used repositories are:
• ArrayExpress hosted by the EBI (http://www.ebi.ac.uk) (Parkinson et al. 2007).
• Gene Expression Omnibus (GEO) at the NCBI (http://www.ncbi.nlm.nih.gov/)
(Barrett et al. 2007).
• Stanford Microarray Database (SMD) at Stanford University (http://
www.stanford.edu) (Demeter et al. 2007).
All repositories have web-forms to query for data, and they offer some dataanalysis features such as normalization, filtering, and clustering. They all offer a
web-based submission process, which is recommended for small to medium-scale
experiments. The choice of repository is a matter of personal preference, but we
recommend that each laboratory uses one repository consistently.
The Microarray Gene Expression Data (MGED) Society (http://www.mged.org)
has developed and promoted standards, recommendations, and tools for sharing
microarray data. The most important in the context of publishing microarray results
is the Minimal Information About a Microarray Experiment (MIAME) standard
V. Mittard-Runte et al.
methods and graphical displays. Within the BioConductor packages there are normalization methods for two-colour arrays as well as for Affymetrix
R
arrays. As a
downside of this flexibility, the system requires a high level of expertise. User interaction is mediated by a command line interface, and complex actions might even
involve programming. Anyway, we would recommend gaining an insight into and
to trying to work with R and BioConductor, as this combination provides maximal
flexibility in the choice of methods and their combination and full control over the
analysis process. Also, novel methods are often first implemented as BioConductor
packages. We refer the reader to Gentleman (Gentleman et al. 2005) for further
reference.
9.4.3 Data Sharing and Public Repositories
Gene expression experiments involve highly complex protocols and produce highvolume data. Compared with a genome sequencing approach the experimental
annotations need to be much more precise, as a good annotation of a functional
genomics experiment also involves environmental conditions. During growth, harvesting protocols, and further laboratory procedures, samples are transformed many
times. Also complex array designs are involved which can carry up to many
hundreds of thousands of features.
The experimental results of a microarray experiment need to be accompanied
with all this meta-information. Otherwise, it is almost impossible to interpret the
resulting data or even reproduce the experiment independently, which is still a difficult and underestimated task. To ensure scientific standards, almost all journals
in functional genomics now require the submission of microarray data to a public
repository prior to publication of results (Ball et al. 2004).
The most frequently used repositories are:
• ArrayExpress hosted by the EBI (http://www.ebi.ac.uk) (Parkinson et al. 2007).
• Gene Expression Omnibus (GEO) at the NCBI (http://www.ncbi.nlm.nih.gov/)
(Barrett et al. 2007).
• Stanford Microarray Database (SMD) at Stanford University (http://
www.stanford.edu) (Demeter et al. 2007).
All repositories have web-forms to query for data, and they offer some dataanalysis features such as normalization, filtering, and clustering. They all offer a
web-based submission process, which is recommended for small to medium-scale
experiments. The choice of repository is a matter of personal preference, but we
recommend that each laboratory uses one repository consistently.
The Microarray Gene Expression Data (MGED) Society (http://www.mged.org)
has developed and promoted standards, recommendations, and tools for sharing
microarray data. The most important in the context of publishing microarray results
is the Minimal Information About a Microarray Experiment (MIAME) standard