PSCBS (version 0.45.1)

estimateKappaByC1Density.PairedPSCBS: Estimate global background in segmented copy numbers

Description

Estimate global background in segmented copy numbers based on the location of peaks in a weighted density estimator of the minor copy number mean levels.

The global background, here called $\kappa$, may have multiple origins where normal contamination is one, but not necessarily the only one.

Assumptions: This estimator assumes that there are segments with C1=0 and C1=1, i.e. some deletions and, typically, some normal segements.

Usage

## S3 method for class 'PairedPSCBS':
estimateKappaByC1Density(this, typeOfWeights=c("dhNbrOfLoci", "sqrt(dhNbrOfLoci)"),
  adjust=1, from=0, minDensity=0.2, ..., verbose=FALSE)

Arguments

typeOfWeights

A character string specifying how weights are calculated.

adjust

A numeric scale factor specifying the size of the bandwidth parameter used by the density estimator.

from

A numeric scalar specifying the lower bound for the support of the estimated density.

minDensity

A non-negative numeric threshold specifying the minimum density a peak should have in order to consider it a peak.

...

Not used.

verbose

See Verbose.

Value

• Returns the background estimate as a numeric scalar.

Algorithm

• Retrieve segment-level minor copy numbers and corresponding weights:
  1. Grabs the segment-level C1 estimates.
  2. Calculate segment weights. The default (typeOfWeights="dhNbrOfLoci") is to use weights proportional to the number of heterozygous SNPs. An alternative (typeOfWeights="sqrt(dhNbrOfLoci)") is to use the square root of those counts.
• Identify subset of regions with C1=0:
  1. Estimates the weighted empirical density function (truncated at zero below). Tuning parameter 'adjust'.
  2. Find the first two peaks (with a density greater than tuning parameter 'minDensity').
  3. Assumes that the two peaks corresponds to C1=0 and C1=1.
  4. Defines threshold Delta0.5 as the center location between these two peaks.
• Estimate the global background signal:
  1. For all segments with C1 < Delta0.5, calculate the weighted median of their C1:s.
  2. Let kappa be the above weighted median. This is the estimated background.

See Also

Instead of calling this method explicitly, it is recommended to use the *estimateKappa() method.