calculateDetection: Calculate detection scores
In beadarray: Quality assessment and low-level analysis for Illumina BeadArray data

Description Usage Arguments Details Value Author(s) Examples

View source: R/ExpressionSetIllumina_calculateDetection.R

Function to calculate detection scores for summarized data if they are not available.

1	calculateDetection(BSData, status=fData(BSData)$Status, negativeLabel="negative")

`BSData`	An ExpressionSetIllumina object
`status`	character vector giving probe types
`negativeLabel`	character giving identifer for negative controls

Detection scores are a measure of whether the probe is showing any specific expression. This function implements Illumina's method for calculating the detection scores for all bead types on a given array. Within an array, Illumina discard negative control bead-types whose summary values are more than three MADs from the median for the negative controls. Illumina then rank the summarized intensity for each other bead-type against the summarized values for the remaining negative control bead-types and calculate a detection p-value 1-R/N, where R is the relative rank of the bead intensity when compared to the $N$ remaining negative controls. Thus, if a particular bead has higher intensity than all the negative controls it will be assigned a value of 0. This calculation is repeated for all arrays.

The function expects the negative controls to be indicated by the Status column in the featureData slot of the ExpressionSetIllumina object. If this is not present the user can supply a status vector with the same length as the number of rows in the ExpressionSetIllumina object.

Matrix of detection scores with the same dimensions as the exprs matrix of BSData. This matrix can be stored in a ExpressionSetIllumina object using the Detection function

Mark Dunning and Andy Lynch

if(require(beadarrayExampleData)){

data(exampleSummaryData)
##By default, the status column of featureData is used

exampleSummaryData.log2 <- channel(exampleSummaryData ,"G")

det <- calculateDetection(exampleSummaryData.log2)

Detection(exampleSummaryData.log2) <- det

##Example of specifying own status vector
  
exampleSummaryData.log2 <- addFeatureData(exampleSummaryData.log2)

pq <- fData(exampleSummaryData.log2)$PROBEQUALITY

det2 <- calculateDetection(exampleSummaryData.log2, status="pq", negativeLabel="No match")
}