Biology Reference
In-Depth Information
FIGURE 2.16
Network representation of the SinI/R system in Bacillus subtilis, indicating important
regulatory relationships.
There are several problems with the use of a Gold Standard for edge confidence esti-
mation. Most importantly, all potential Gold Standards, nomatter howwell curated, are
incomplete, and may contain errors ( Cusick et al. ,2008 ). After all, if the Gold Standard
was complete we would have no need to build integrated networks in the first place.
Different Gold Standards produce different weightings, because they contain different
data. Recent research has investigated the use of problem relevance in weighting edges
( James et al. ,2009 ) and the assessment of edge probability without the use of a Gold
Standard ( Weile et al. ,2012 ), but such approaches are still not widely used.
There are many possible sources of data for the construction of integrated inter-
action networks ( interactomes ). Up-to-date information about publically available
databases is available in the annual Nucleic Acids Research Database issue, pub-
lished in January of each year. The 2012 issue details 1380 databases ( Galperin
and Fern´ndez-Su´rez, 2012 ). Not all of these databases contain data relevant to
microbes, and those that do generally cover a limited range of species. Although
there is a relatively large amount of data for model species such as S. cerevisiae ,
the amount of high-throughput data available for less well-studied species may be
limited. Several of the major databases include data from multiple single-data-type
sources. BioGrid 13 ( Stark et al. , 2011 ), for example, is widely used, incorporates data
13 http://thebiogrid.org/ .
Search WWH ::




Custom Search