RCytoscape- Non-square adjacency matrix

[Chirag Gupta]

Hi

I have a non square adjacency matrix with rows representing genes and columns as Transcription Factors in R. Each cell value is a score of either 0 or 1. I want to view this in cytoscape. All the available functions ask for a square matrix. I am stuck up in this situation for days now. 

Any help is appreciated.

Thanks. 

[Tim Kacprowski]

hello,

that sounds like you are dealing with a bi-partite graph represented by that matrix.

you can, e.g.,  represent this graph with an usual square adjacency matrix, as well.

cheers

tim

________________________________________

Tim Kacprowski

http://timkacprowski.wordpress.com/

Molecular Networks in Medical Bioinformatics

Max Planck Institute for Informatics

Campus E1.4, 66123 Saarbrücken, Germany

________________________________________

--
You received this message because you are subscribed to the Google Groups "cytoscape-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cytoscape-disc...@googlegroups.com.
To post to this group, send email to cytoscap...@googlegroups.com.
Visit this group at http://groups.google.com/group/cytoscape-discuss.
For more options, visit https://groups.google.com/groups/opt_out.

[Paul Shannon]

Can you give me a small example -- as small as it can be but sufficient to demonstrate your difficulty?

I'd be glad then to help out.

- Paul

[Chirag Gupta]

Lets say

> y <- matrix(rnorm(50), 10, 5, dimnames=list(paste("g", 1:10, sep=""), paste("g", 1:5, sep="")))

> y

            g1          g2         g3          g4          g5

g1   0.3074333 -0.27050232 -0.3779216 -2.49669225  0.09672238

g2  -1.4595533  1.33653556 -0.6434152  0.69457288  0.12146522

g3   0.7390419 -0.75854672  0.6254186  0.33169744 -1.01592252

g4   0.4479730 -0.55661946  0.2709066 -0.04312139  0.83234461

g5   0.6056811 -0.68056140  0.3908853  0.65742583 -1.64629457

g6  -0.4503781 -0.01651466  0.7885615 -0.76685821  0.74624990

g7   1.3214891 -0.78996628 -0.6776208 -1.23106417 -1.43658047

g8   0.3922689  1.37774633  0.5571683 -1.05571200  0.32167027

g9  -0.1545101 -0.61269863  0.5805345 -1.01307431 -0.45720747

g10 -0.2896233  1.16499032 -0.5502498  0.25729953  0.38061157

> a.matrix=ifelse(c(y)>0,1,0)

> dim(a.matrix)<-dim(y)

> a.matrix

      [,1] [,2] [,3] [,4] [,5]

 [1,]    1    0    0    0    1

 [2,]    0    1    0    1    1

 [3,]    1    0    1    1    0

 [4,]    1    0    1    0    1

 [5,]    1    0    1    1    0

 [6,]    0    0    1    0    1

 [7,]    1    0    0    0    0

 [8,]    1    1    1    0    1

 [9,]    0    0    1    0    0

[10,]    0    1    0    1    1

Notice that this is a non square matrix and I want to view this in cytoscape. Now I want to flatten it, ie convert it into a .sif format. (it would be better if I could get three columns with one of the columns having the corresponding '0' or '1', and the other two specifying rows and columns).

eg:

g1 g1 1

g1 g2 0

g1 g3 0

g1 g4 0

g1 g5 1

g2 g1 0

g2 g2 1

etc...

 

Also, this is not a bipartite graph (some of the genes might be same in rows and columns). It can be represented like that but thats not the goal.

Thanks for help!  

[Paul Shannon]

Hi Chirag,

My solution, shown below, starts with your matrix, adds empty columns to make it square, constructs a graphAM using the matrix, converts to a graphNEL, and then displays the result using RCytoscape:

y <- matrix(rnorm(50), 10, 5, dimnames=list(paste("g", 1:10, sep=""), paste("g", 1:5, sep="")))

x <- matrix(0, 10, 5, dimnames=list(paste("g", 1:10, sep=""), paste("g", 6:10, sep="")))
xy <- cbind(y,x)
xy[xy <=0 ] <- 0
xy[xy >0 ] <- 1
library(RCytoscape)
g1 <- graphAM(xy, edgemode="directed")
g2 <- as(g1, "graph")
cw <- new.CytoscapeWindow("from adjacency matrix", g2, overwrite=TRUE)
displayGraph(cw)

setNodeLabelRule(cw, "label")

redraw(cw)

With this (rather hasty) result.  Is this helpful?

 - Paul

-- 
Thanks and regards,
Chirag Gupta
+1479-799-7136

[Tim Kacprowski]

i see.

following some quick thoughts, you could do:

cat( paste( apply( which(a.matrix == 1, arr.ind = T), 1,

  function (x) {

paste(x[1], x[2], a.matrix[x[1],x[2]],

 sep ='\t')

  }

),

collapse = '\n'))

but there may also be some more elegenat formulation if you think about it for a while...

cheers

tim

________________________________________

Tim Kacprowski

http://timkacprowski.wordpress.com/

Molecular Networks in Medical Bioinformatics

Max Planck Institute for Informatics

Campus E1.4, 66123 Saarbrücken, Germany

________________________________________

--

[Tim Kacprowski]

to get 'g1' instead of 1, replace x[1] by rownames(y)[x[1]] in the inner paste...

________________________________________

Tim Kacprowski

http://timkacprowski.wordpress.com/

Molecular Networks in Medical Bioinformatics

Max Planck Institute for Informatics

Campus E1.4, 66123 Saarbrücken, Germany

________________________________________