rqubic (version 1.18.0)

quBicluster: Qualitative Biclustering

Description

The function takes seeds and quantileDiscretized ExpressionSet as input, biclusters the data and returns an object holding biclusters. Users may control the report number of clusters, tolerance of incoherent genes (or conditions), as well as the filtering of redundant clusters.

Usage

quBicluster(seeds, eset, report.no = 100L, tolerance = 0.95, filter.proportion = 1)

Arguments

seeds

An object of the S3-class rqubicSeeds, representing seeds generated from the quantileDiscretized expression data

eset

Discretized expression data

report.no

Number of biclusters that should be reported. Detected biclusters are ranked by the S-score, which is defined by the product of gene counts and sample counts. They are ordered and the top ones are reported.

tolerance

Percentage of tolerated incoherent samples, 0.95 by default

filter.proportion

Proportion of a cluster, over which the cluster is considered as redudant. Each bicluster is compared to all better ranking biclusters, and the overlapping proportion is measured by the proportion of the product of overlapping samples and overlapping genes, to the product samples and genes. If the proportion is larger than the given threshold, the block will be considered redundant and therefore not reported. Setting the threshold to 1 (default) does not perform any filtering.

Value

An object of the QUBICBiclusterSet-class, holding all biclusters.

Details

The function calls a C routine to perform the biclustering. Currently the routine returns blocks with fewer samples specified by the minimum column number, due to the set of tolerance values. This might be changed in the fewer versions.

References

Li et al. (2009) QUBIC: a qualitative biclustering algorithm for analyses of gene expression data Nucleic Acids Research 37:e101

See Also

quantileDiscretize and generateSeeds

Examples

data(sample.ExpressionSet, package="Biobase")
rqubic.example.discret <- quantileDiscretize(sample.ExpressionSet, rank=2L)

rqubic.example.sel.seeds <- generateSeeds(rqubic.example.discret, minColWidth=2L)

rqubic.example.blocks <- quBicluster(rqubic.example.sel.seeds,
    rqubic.example.discret,
    report.no=200L,
    filter.proportion=0.1)

## print features in each bicluster
BCfeatures(rqubic.example.blocks)