Solutions - Bayesian Networks in R: With Applications in Systems Biology - page 139

> for (i in 1:50){
> DBNGeneNet.50edges.mat[DBNGeneNet.edges$
node2[i],
+ DBNGeneNet.edges$node1[i]] = 1
> }#FOR
> DBNsimone.50edges.mat = DBNsimone.50edges.net$A
> DBNG1DBN.50edges.mat =
+ as.numeric(DBNG1DBN < G1DBN.thres.top50)
> sum(DBNG1DBN.50edges.mat, na.rm = TRUE)
> sum(which(DBNlasso.50edges.mat == 1) %in%
+ which(DBNGeneNet.50edges.mat == 1))
> sum(which(DBNlasso.50edges.mat == 1) %in%
+ which(DBNG1DBN.50edges.mat == 1))
> sum(which(DBNlasso.50edges.mat == 1) %in%
+ which(DBNsimone.50edges.mat == 1))
(d) Different dimension reduction procedures select significantly different sets of
arcs. It is likely that the various approaches have different power in identifying
the dependencies present in the data and therefore complement each other.

Exercises of Chap. 4

4.1 Apply the junction tree algorithm to the validated network structure from
Sachs et al. ( 2005 ), and draw the resulting undirected triangulated graph.

PKC plcg

P38 PIP2

PKA PIP3

PKA PIP3

P38 praf pjnk PIP2

p44.42 PKC plcg

pmek

pakts
473

pmek pjnk

p44.42

praf

pakts
473

4.2 Consider the Sachs et al. ( 2005 )datausedinSect. 4.2 .
(a) Perform parameter learning with the bn.fit function from bnlearn and
the validated network structure. How do the maximum likelihood estimates