genSimData: Simulate an omics dataset with biclusters
In jonalim/mfBiclust: Pipelines for Bi-Clustering Using Matrix Factorization

Description Usage Arguments Details Value Examples

Most models of omics datasets can be built using various combinations of effects in this function. Effects are applied in this order:

Constant background value
Base row values (overrides constant background value)
Gaussian noise
Constant value across all biclusters (overrides all previous effects)
Additive bicluster-specific values
Additive bicluster-and-column-specific values
Multiplicative bicluster-and-column-specific values
Additive bicluster-and-row-specific values
Multiplicative bicluster-and-row-specific values
Shuffling of rows and columns

genSimData(n = 1, clusterHeight = 20, clusterWidth = 20, dimx = 80,
  dimy = 80, overlapRows = 0, overlapCols = 0, bgConst = 0,
  bgNorm = 0, biclusterConstant = NULL, biclusterShift = 0,
  rowBase = 0, rowShift = 0, rowScale = NULL, colShift = 0,
  colScale = NULL, shuffle = TRUE, file = "")

`n`	the number of biclusters
`clusterHeight`	the number of rows in each bicluster
`clusterWidth`	the number of columns in each bicluster
`dimx`	the number of rows in the dataset
`dimy`	the number of columns in the dataset
`overlapRows`	the number of rows that biclusters should share
`overlapCols`	the number of columns that biclusters should share
`bgConst`	initial value of the entire matrix
`bgNorm`	standard deviation of Gaussian noise
`biclusterConstant`	if not NULL, sets the value of every element in every bicluster, after creating the background
`biclusterShift`	standard deviation of bicluster-specific additive values
`rowBase`	standard deviaton of base row values
`rowShift`	standard deviation of bicluster-and-row-specific shifts
`rowScale`	standard deviation of bicluster-and-row-specific scaling coefficients
`colShift`	standard deviation of bicluster-and-column-specific shifts
`colScale`	standard deviation of bicluster-and-column-specific scaling coefficients
`shuffle`	if FALSE, biclusters will be deterministically arranged along the diagonal of the matrix
`file`	an optional string giving a prefix for .png and .csv files where the generated matrix should be written

Besides constant values across all biclusters (biclusterConstant), all effects are sampled from a normal distribution centered on 0 and with standard deviation given by the respective parameter.

A numeric matrix containing n biclusters created by the given parameters.

# No background, sixteen non-overlapping 5x5 biclusters. Elements in
# biclusters have value 5; all other elements are 0.
mat <- genSimData(n = 16, biclusterConstant = 5,
 clusterHeight = 5, clusterWidth = 5, shuffle = FALSE)

# No background, two 5x5 biclusters causing row-shift effects
mat <- genSimData(n = 2, rowBase = 1, rowShift = 1,
 clusterHeight = 5, clusterWidth = 5, shuffle = FALSE)
matrixHeatmap(mat)

# Three 5x5 biclusters, where the first and second biclusters overlap and the
# second and third biclusters overlap. Both overlap regions are 4 rows by 2
# columns.
mat <- genSimData(n = 3, rowBase = 1, rowShift = 1, overlapRows = 4,
 overlapCols = 2, clusterHeight = 5, clusterWidth = 5, shuffle = FALSE)
matrixHeatmap(mat)

# One 10x10 plaid bicluster
mat <- genSimData(n = 3, biclusterShift = 1, rowShift = 1, colShift = 1,
 clusterHeight = 10, clusterWidth = 10, shuffle = FALSE)
 matrixHeatmap(mat)