beadarray (version 2.22.2)

BASHCompact: BASH - Compact Defect Analysis

Description

Creates a list of probes marked as being in compact defects.

Usage

BASHCompact(inten, probeIDs, neighbours = NULL, wts=1, n=3, maxiter = 10, cutoff = 8,  cinvasions = 10, outlierFun=illuminaOutlierMethod, ...)

Arguments

inten

the (transformed) intensities associated with beads on an array.

probeIDs

the probe identities associated with those beads (i.e. indicating which beads are of which sort)

neighbours

A Neighbours matrix such as that generated by generateNeighbours. Compulsory - the function will no longer generate this for you.

wts

weights indicating any beads to be masked in calculations.

n

Specify a cut-off for outliers (e.g. as n median absolute deviations (MADs) from the median). The default value is 3

maxiter

Integer - Maximum number of iterations.

cutoff

Integer - Size a cluster must be to be labelled a compact defect.

cinvasions

Integer - Number of invasions used when closing the image.

outlierFun

the choice of outlier calling function to use.

...

Additional arguments to be passed to outmeth.

Value

A vector consisting of the BeadIDs of beads labelled as compact defects.

Details

BASHCompact finds "compact defects" on an array. A compact defect is defined as a large connected cluster of outliers.

This function first finds the outliers on an array. This is done via the user's choice of function (e.g. illuminaOutlierMethod or squeezedVarOutlierMethod.

Next, using the Neighbours matrix and a Flood Fill algorithm, it determines which beads are in large connected clusters of outliers (of size larger than cutoff). These beads are then temporarily removed and the process repeated with the remaining beads. The repetition continues until either no large clusters of outliers remain, or until we have repeated the process maxiter times (and in this case, a warning will be given). In this way, we obtain a list of defective probes.

Finally, we "close" the image, to fill in small gaps in the defect image. This consists of a "dilation" and an "erosion". In the dilation, we expand the defect image, by adding beads adjacent to defective beads into the defect image. This is repeated cinvasions times. In the erosion, we contract the defect image, by removing beads adjacent to non-defective beads from the defect image. (Erosion of the defect image is equivalent to a dilation of the non-defective image.)

References

J. M. Cairns, M. J. Dunning, M. E. Ritchie, R. Russell, and A. G. Lynch (2008). BASH: a tool for managing BeadArray spatial artefacts. Bioinformatics 15; 24(24)

See Also

BASH, generateNeighbours

Examples

## Not run: 
# 
# if(require(beadarrayExampleData)){
# 
# data(exampleBLData)
# o <- BASHCompact(getBeadData(exampleBLData,array=1,what="Grn"), getBeadData(exampleBLData,array=1,what="ProbeIDs"))
# o <- BASHCompact(getBeadData(exampleBLData,array=1,what="Grn"), getBeadData(exampleBLData,array=1,what="ProbeIDs"), cinvasions = 10) ##increased no of closure invasions
# o <- BASHCompact(getBeadData(exampleBLData,array=1,what="Grn"), getBeadData(exampleBLData,array=1,what="ProbeIDs"), cutoff = 12) ##only larger defects will be found with this setting
# 
# }
# 
# ## End(Not run)