Biology Reference
In-Depth Information
is incomplete. For example, the widely used Arabidopsis Affymetrix ATH1
GeneChip covers only 63% of the genes found in A. thaliana, potentially
resulting in large number of false negatives; (ii) Incomplete coverage of
tissues and developmental stages can result in both false negatives (e.g.,
co-expression relationships of pollen specific genes are unattainable if no
tissue containing pollen is present in the microarray dataset) and false
positives (e.g., difficulties in distinguishing pollen- and ovule-specific genes
if only flowers are measured); (iii) Low spatio-temporal resolution of gene
expression can fail to separate co-regulated processes that can result in false
positives, as shown in above example, where lignin biosynthesis was
co-regulated with cellulose biosynthesis. While it is plausible that lignin
biosynthesis is coupled with cellulose biosynthesis during secondary cell
wall formation, the analysis failed to separate these two distinctive processes.
Still, these shortcomings can be alleviated by performing cross-species
co-expression analysis. Lignin is present in mono- and dicotyledons and is
produced by similar enzymes in both kingdoms (discussed in previous para-
graphs). Several studies have also shown that co-expressed relationships are
conserved across species and even kingdoms (
Bergmann et al., 2004; Stuart
et al., 2003
). Therefore, by applying co-expression analysis to lignin biosyn-
thesis in multiple species and extracting the commonalities between the results,
it is possible to highlight recurring genes (i.e., enrich true positives). Another
advantage of cross-species comparison is detection of functional homologs.
A major goal of applied biology is to transfer the knowledge obtained in a
model organism (donor), such as Arabidopsis, to other species (acceptors),
which have greater economic and nutritional importance. Once the function of
a gene product in the knowledge donor has been proven experimentally,
uncovering the identity of the functional equivalent in an acceptor species is,
however, not trivial. As plant genomes characteristically contain large gene
families, sequence comparison of a gene from the knowledge donor to the
genome of the acceptor can return a large list of possible candidate genes.
While several of those candidates may perform the same molecular function,
they are not necessarily equivalent participants in the biological process of
interest. Intuitively, a functional homolog should be present when the relevant
biological process occurs. Thus, functional homologs from different species
should be reflected in conserved co-expression patterns.
B. CROSS-SPECIES CO-EXPRESSION ANALYSIS
PlaNet database (
www.aranet.mpimp-golm.mpg.de
), which is loaded with
microarray data for Arabidopsis, barley (Hordeum vulgare), rice, Medicago,
poplar (Populus spp.), wheat (T. aestivum) and soybean (Glycine max.), and
