filter.biclust: Filter biclusters by overlap and quantity
In jonalim/mfBiclust: Pipelines for Bi-Clustering Using Matrix Factorization

Description Usage Arguments Value Examples

First all biclusters encompassing the whole matrix are removed. Next, biclusters are selected in descending order of size, skipping any biclusters that overlap excessively with already- selected biclusters. The remaining k^\prime biclusters are returned in membership-matrix format, along with a matrix giving the union of the retained biclusters and a logical vector telling which biclusters were retained. See clusteredSamples() for a description of the membership matrix format.

1 2	filter.biclust(rowxBicluster, biclusterxCol, max = NULL, overlap = 0.25)

`rowxBicluster`	a row-bicluster membership matrix
`biclusterxCol`	a bicluster-column membership matrix
`max`	the maximum number of biclusters to keep
`overlap`	the maximum portion of a bicluster that may overlap with another bicluster. Passed directly to `overlap()`

A list of the format:

RowxBicluster: A row-bicluster membership matrix A_{m,k^\prime}, where k^\prime is the number of retained biclusters
BiclusterxCol: A bicluster-column membership matrix B_{k^\prime,n}
biclustered: A binary numeric matrix giving the union of all retained biclusters
chosen: A logical vector of length k

bce <- BiclusterExperiment(yeast_benchmark[[1]])
bce <- addStrat(bce, k = 4, method = "als-nmf")
bcs <- getStrat(bce, 1)
filtered <- filter.biclust(clusteredFeatures(bcs), clusteredSamples(bcs),
  overlap = 0.25)
filtered$chosen
# The second bicluster was excluded because .31 of its matrix elements
# overlapped with the first bicluster.