Biomedical Engineering Reference
In-Depth Information
modulated by the protein target of a small molecule. In conclusion, yeast will con-
tinue to be a valuable tool for studying the mechanisms of action and resistance to
drugs and in designing new drugs. The value of a genomic approach to an understand-
ing of fundamental biological processes is connected to the simultaneous examination
of expression patterns of all genes for branch-point enzymes. Similarly, the defini-
tion of the expression profile induced by the treatment of yeast cells with a selected
drug or molecule allows for the identification of genes mediating the drug response.
Nevertheless, developing the complete transcriptional profile of a cell might be as
complicated as looking for a needle in a haystack. Indeed, a huge number of genes can
change in their expression level, thus making difficult the disclosure of a biologically
meaningful portrait. Deep knowledge of the central metabolism and the use of bioin-
formatics tools for data mining, hierarchical organization, and clustering analysis is
thus required [85,86]. Aiming at a simplification of the results and at other tasks,
many different analyses have been proposed since the high-throughput microarray
technique begin to gain interest. The expression levels associated with each gene can
be analyzed using a supervised or an unsupervised approach. The first is applicable
when further information on the profiles is available. As an example, by attributing
a characteristic (e.g., diseased/normal, treated/untreated) to a sample it is possible
to build a classifier able to predict the sample status from the expression profile,
through the identification of specific signatures. Unsupervised data analysis consists
of grouping together objects (be they genes or samples) by clustering the expression
profiles and allowing for the identification of common features [87]. Clustering is
not a new technique, as it was developed initially for phylogenetic analyses. Many
different ways to calculate distances (and thus similarities/differences) describing
transcriptional profile differences are available [86,88,89], and several tools have
been developed with this aim. The combination of sample and gene clustering allows
for the identification of groups of genes whose expression can distinguish one group
of samples (a subcluster) from another sample subcluster. Other approaches have
been developed aiming at the disclosure of a molecular profile or biological effect
induced by the perturbation studied. In the early days, procedures were based on
the utilization of patterns of genes grouped as belonging to the same domain or
pathway. The principal domains defined by the Gene Ontology Project [90] are:
cellular component
, the parts of a cell or its extracellular environment;
molecular
function
, the elemental activities of a gene product at the molecular level, such as
binding or catalysis; and
biological process
, operations or sets of molecular events
pertinent to the functioning of integrated living units (i.e., cells, tissues, organs, and
organisms). The Gene Ontology Consortium, in a collaboration among three model
organism databases (FlyBase for
Drosophila
[91], SGD for
S. cerevisiae
[92], and
MGD for mouse [93]), developed controlled vocabularies (ontologies) that describe
gene products in terms of their associated domain. Also, pathway databases have
been annotated as a result of information gained from the different research domains.
The combination of genetic, functional, and proteomic data allowed for the gen-
eration of lists of functionally or metabolically related genes. The main pathways
data sets curated for
S. cerevisiae
are (1) KEGG [94], a collection of manually
drawn pathway maps representing the molecular interaction and reaction networks
