The Lotus japonicus Genome

(Steven Felgate) #1

display has several features such as zoom in/
zoom out and move options. Each node with its
connection strength and prediction approaches is
shown by line width and color, respectively. All
circles represent genes (TFs are in orange and
putative target genes in open circles) and are
clickable to link them to their probe set ID, gene
ID, tentative annotation and GO term on the right
panel (Fig.17.4b). Clicking on the edge displays
the connection strength value calculated by the
corresponding GRN prediction algorithm and a
link to show the gene expression profiles for the
related gene pair in the LjGEA Web server.


17.5.3 Analysis

Features of GRN analysis include comparison
and integration of multiple networks predicted by
different algorithms. Users can choose to overlay
multiple GRNs and construct a composite net-
work in order to highlight similarities and differ-
ences between predictive algorithms (Fig.17.4c).
Another feature is“sub-network query.”Usually,
GRNs are too large and too complex to be dis-
played or analyzed. LegumeGRN allows users to
extract sub-networks by showing only immediate
connections of a specified gene list or by selecting

Fig. 17.4 The LegumeGRN Web server.aHomepage of
the Web server with tabs to select species, tools, and gene
network browser on thetop. This homepage displays the
four parts that will be used to build GRN: list of probe
sets/gene input, general options, selection of transcrip-
tomic data and predictive algorithms.bExample of a
GRN visualized using the Web-based GRN viewer. This
visualization module consists of a graphical output on the


left panel, where TFs (orange nodes) are connected to
putative target genes (gray nodes) and an annotation panel
on theright side.cVisualization of connections from a
GRN built using two different algorithms: connections
from relevance network are identified bypurple linesand
from the GGM algorithm withgray lines. This network
highlights similarities between connections predicted
from the two different predictive algorithms

196 J. Verdier et al.

Free download pdf