createBeadSummaryData {beadarray} | R Documentation |
Produce bead averages for each bead type used in an experiment on a specified set of strips/arrays.
createBeadSummaryData(BLData, log=FALSE, imagesPerArray = 1, what="G", probes = NULL, arrays=NULL, method="illumina", n=3, trim=0.05)
BLData |
BeadLevelList |
log |
if TRUE then summarise the log2 intensities of each bead |
imagesPerArray |
Specifies how many images (strips) there are per array. Normally 1 for a SAM and 1 or 2 for a BeadChip. The images (strips) from the same array will be combined so that each column in the output represents a sample |
what |
character string specifying which intensities/values
to summarise. See getArrayData for a list of possibilities. |
probes |
Specify particular probes to summarise. If left NULL then all the probes on the first array are used. |
arrays |
integer (scalar or vector) specifying the strips/arrays
to summarise. If NULL , then all strips/arrays are summarised. |
method |
chracter string specifying the summarisation method to
use. Options are "illumina" , "mean" , "median" ,
"trim" and "winsorise" . |
n |
numeric value defining a cut-off for the number of
median absolute deviations (MADs) from the median to use for
determining outliers. The default value is 3. Used when
method="illumina" |
trim |
fraction of intensities to remove from the bead summary
calculations when method="trim" , or the fraction of
intensities to set to the trim and 1-trim percentile
intensities when method="winsorize" . Default value is 0.05. |
To summarise the raw data using the default method used by Illumina
(method="illumina"
) we first remove outliers for each bead
type on each array. Outliers are beads which have an intensity greater
than 3 median absolute deviations (MADs) from the bead median intensity on
the original (un-logged) scale. The n
argument can be changed to
remove beads with intensity n
MADs above or below the median.
With outliers removed, the average (mean) intensities of the remaining
beads are calculated along with the standard error and number of beads.
Other summarisation options are also available. When
method="mean"
, the average and standard error of all beads
for a given bead type is calculated on each array. This would be
appropriate if the scanner has been set up to exclude outlier beads
from the bead level .txt or .csv files.
When method="median"
, the middle value is returned along with the
median absolute deviation (rather than standard error) for each bead type.
When method="trim"
, the trimmed mean and standard error are
calculated and for method="winsorize"
the winsorised mean and
standard error are returned.
By setting the log
argument to TRUE
, we calculate outliers and
summary values on the log2-scale.
If there are any NAs or Inf values, they are ignored.
Objects which are created separately by 'createBeadSummaryData' may
be joined using the combine
function.
An ExpressionSetIllumina
object (or NChannelSet
object for two-colour data, when what="RG"
) in which all
components are matrices with number of rows equal to the number of bead
types for the experiment and number of columns equal to the number
of arrays.
Mark Dunning and Mike Smith
#produce bead summaries for each array data(BLData) BSData = createBeadSummaryData(BLData, log=TRUE, what="G") dim(BSData)