ppa.iterate: The Ping-Pong Algorithm

Description

Perform PPA on two (normalized) input matrices.

Usage

"ppa.iterate"(normed.data, ...)

Arguments

normed.data

The normalized input matrices. A list of four matrices, usually coming from the ppa.normalize function.

...

Additional arguments, see details below.

Value

link{isa} for a complete description.

Details

ppa.iterate performs the PPA iteration on the specified input matrices using given input seeds. It can be called as

    ppa.iterate(normed.data, row1.seeds, col1.seeds,
                row2.seeds, col2.seeds,
                thr.row1, thr.col=thr.row1, thr.row2=thr.row1,
		direction="updown",
		convegence=c("cor"), cor.limit=0.9,
		oscillation=FALSE, maxiter=100)

where the arguments are:

normed.data: The normalized data, a list of four matrices with the appropriate size. They are usually coming from the output of the ppa.normalize function.
row1.seeds: The row seed vectors for the first matrix. At least one of the four possible seed vectors must be present and they will be concatenated, after doing the suitable half-iterations.
col1.seeds: The column seed vectors for the first matrix. At least one of the four possible seed vectors must be present and they will be concatenated, after doing the suitable half-iterations.
row2.seeds: The row seed vectors for the second matrix. At least one of the four possible seed vectors must be present and they will be concatenated, after doing the suitable half-iterations.
col2.seeds: The column seed vectors for the second matrix. At least one of the four possible seed vectors must be present and they will be concatenated, after doing the suitable half-iterations.
thr.row1: Numeric scalar or vector giving the threshold parameter for the rows of the first matrix. Higher values indicate a more stringent threshold and the result comodules will contain less rows for the first matrix on average. The threshold is measured by the number of standard deviations from the mean, over the values of the first row vector. If it is a vector then it must contain an entry for each seed.
thr.col: Numeric scalar or vector giving the threshold parameter for the columns of both matrices. The analogue of thr.row1.
thr.row2: Numeric scalar or vector, the threshold parameter(s) for the rows of the second matrix. See thr.row1 for details.
direction: Character vector of length four, one for each matrix multiplication performed during a PPA iteration. It specifies whether we are interested in rows/columns that are higher (‘up’) than average, lower than average (‘down’), or both (‘updown’). The first and the second entry both corresponds to the common column dimension of the two matrices, so they should be equal, otherwise a warning is given.
convergence: Character scalar, the convergence criteria for the PPA iteration. If it is ‘cor’, then convergence is measured based on high Pearson correlation (see the cor.limit argument below) of the subsequent row and column vectors. Currently this is the only option implemented.
cor.limit: The correlation limit for convergence if the ‘cor’ method is used.
oscillation: Logical scalar, whether to look for oscillating seeds. Usually there are not too many oscillating seeds, so it is safe to leave this on FALSE.
maxiter: The maximum number of iterations allowed.

References

Kutalik Z, Bergmann S, Beckmann, J: A modular approach for integrative analysis of large-scale gene-expression and drug-response data Nat Biotechnol 2008 May; 26(5) 531-9.

Examples

Run this code

## A basic PPA workflow with a single threshold combination
## In-silico data
set.seed(1)
insili <- ppa.in.silico()

## Random seeds
seeds <- generate.seeds(length=nrow(insili[[1]]), count=100)

## Normalize input matrices
nm <- ppa.normalize(insili[1:2])

## Do PPA
ppares <- ppa.iterate(nm, row1.seeds=seeds, thr.row1=1, direction="up")

## Eliminate duplicates
ppares <- ppa.unique(nm, ppares)

## Fitler out not robust ones
ppares <- ppa.filter.robust(insili[1:2], nm, ppares)

## Print the sizes of the co-modules
cbind(colSums(ppares$rows1 != 0), colSums(ppares$rows1 != 0),
      colSums(ppares$columns != 0))

## Plot the original data and the first three modules found
if (interactive()) {
  layout(rbind(1:2,3:4))
  image(insili[[1]], main="In silico data 1")
  image(outer(ppares$rows1[,1],ppares$columns[,1])+
        outer(ppares$rows1[,2],ppares$columns[,2])+
        outer(ppares$rows1[,3],ppares$columns[,3]), main="PPA co-modules 2")
  image(insili[[2]], main="In silico data 2")
  image(outer(ppares$rows2[,1],ppares$columns[,1])+
        outer(ppares$rows2[,2],ppares$columns[,2])+
        outer(ppares$rows2[,3],ppares$columns[,3]), main="PPA co-modules 2")
}

Run the code above in your browser using DataCamp Workspace