Object Oriented Microarray Library: Channel

The channel module provides the definition of the channel class. Channel objects are typically extracted from the raw data in a complete.channel object; see the section on extractors and transformers for information on how this is accomplished. See the bottom of the page for an example of how the class can be used.

Requirements

This class uses the complete.channel and utility packages, which must be loaded in order for the routines in this package to work correctly.

An extractor is a function that takes a complete.channel object as its input (along with an optional auxiliary list of parameters) and produces a channel object as its output. A transformer is a function that takes a channel object as its input (along with an optional auxiliary list of parameters) and produces another channel object as its output. The library supplies several examples of extractors and transformers for use with microarray data.

bkgd.extractor

Extracts the vector of background measurements.

vol.extractor

Extracts the vector of raw volume measurements.

svol.extractor

Extracts the ArrayVision-computed vector of background-corrected volumes, which replaces negative values by zero.

negsvol.extractor

Extracts the vector of raw background-corrected volumes, which retains possible negative values.

sd.extractor

Extracts the vector of pixel-wise standard deviation estimates; this is not guaranteed to work, because they may not have been saved by the ArrayVision operator.

sn.extractor

Extracts the vector of signal-to-noise ratios; this is not guaranteed to work, because they may not have been saved by the ArrayVision operator.

subtract.transform(object, value)

Subtracts the given value from all elements of the object.

threshold.transform(object, level)

Replaces all data values in the object that are below the given level by that value. The default threshold level is zero.

normalize.transform(object, n)

Divides all data values in the object by the given normalization factor n. The default value of n is zero. In this case, n is replaced by one one-thousandth of the 75th percentile. (In other words, after applying the default normalization transform, the 75th percentile of the new data set has been set equal to 1000.)

median.expressed.transform(object)

Normalizes the data multiplicatively by rescaling the median of expressed genes (values above the zero background level) to be 1000.

subset.normalize.transform(object, selector)

Normalizes the object using a subset of the data. The selector argument should be a logical vector equal to the number of genes. All values in the channel are normalized by setting the median of the selected values equal to 1000.

subset.mean.normalize.transform(object, selector)

Normalizes the object using a subset of the data. The selector argument should be a logical vector equal to the number of genes. All values in the channel are normalized by setting the mean of the selected values equal to 1000.

log.transform(object)

Applies a log-transform (base 2) to the data in the object.

Attributes

parent

A character string describing the complete.channel object from which this object was constructed.

name

A character string describing the meaning of the measurements stored in this object.

type

An object of the channel.type object describing the kind of microarray experiment that this data came from.

x

The actual vector of measurements; i.e., the whole reason this object exists.

Methods

standard.channel(cc, ef, ep, nf, np, tf, tp, lf, lp)

This is the standard method for constructing a channel object. Only the first argument is required; it must be a complete.channel. The default values for the remaining arguments are

• ef = negsvol.extractor, ep=0
• nf = normalize.transform, np=0
• tf = threshold.transform, tp=25
• lf = log.transform, lf=0

The functions passed in as arguments are applied in this order to the data. Using the default values, we first compute the quantity vol - bkgd, normalize by setting the 75th percentile to 1000, threshold at 25, and then log-transform. You can change just the threshold by adjusting the tp parameter.

print(object, ...)

Print all the data on the object.

summary(object, ...)

Write out a summary of the object.

plot(object, ...)

This method produces two graphical representations of the object. The first is a scatter plot of the measurement values against their index in the vector c, which serves as a surrogate for the position on the microarray. The second is a histogram of the data values.

image(object, ...)

This method produces a two-dimensional "cartoon" image of the measurement values, with the position in the cartoon corresponding to the two-dimensional arrangement of spots on the actual microarray.

Description

An object of the channel class represents a single kind of measurement performed at all spots of a complete.channel. These objects are essentially just vectors of data, with length equal to the number of spots on the microarray, with some extra metadata attached.

  x <- f.load.clontech.file('/huff/05082000a-4.txt', 'wild')

  xv <- log.transform(vol.extractor(x))
  summary(xv)
  plot(xv)
  image(xv)

  b <- negsvol.extractor(x,0)
  n <- normalize.transform(b,0)
  t <- threshold.transform(n, t=25)
  w <- log.transform(t,0)
  plot(w)
  image(w)

  q <- standard.channel(x)
  summary(q)
  plot(q)
  image(q)