ACGHsegment

Segments Array CGH data with the Circular Binary Segmentation (CBS) algorithm and produces hard or soft copy number aberration calls.

The 'geneAnnotation' input file must contain columns Chr, start, and end separately and the chromosomal numbers must be without the string 'chr' in front of the chromosome identifier. Valid chromosome identifiers are [1,2,..22, X, Y]. The actual names of the column names of the identifiers in the 'geneAnnotation' file are user definable.

Two output modes exist. The first one (hard) calls the copy number of a probe aberrated if is more than the median plus two standard deviations apart from the mean of the intensities of all the samples under scrutiny. In the second (soft) copy number aberration calls are made probabilistically with either the CGHcall or the FastCall (part of the TASSO package) algorithm. Please note that FastCall only works on 32bit systems.

Several plots can be produced based on user parameters. Example images can be found here.

The outputs are the dependant on the analysis procedure. The following matrices can be output: All segments with their respective copy number aberration calls, thresholds for significant copy number aberrations, and segment call probabilities.

Version	2.0
Bundle	microarray
Categories	Agilent Copy Number Analysis
Authors	Riku Louhimo (Riku.Louhimo@Helsinki.FI)
Issue tracker	View/Report issues
Requires	R ; limma (R-bioconductor) ; DNAcopy (R-bioconductor) ; CGHcall (R-bioconductor) ; TASSO (R-package)
Source files	component.xml ACGHsegment.r
Usage	Example with default values ACGHsegment( caseChan = CSV, geneAnnotation = AnnotationTable, CGHCallMaxnumseg = 100, CGHCallPrior = "auto", CGHCallRobustsig = true, alpha = 0.01, bpEndCol = "end", bpStartCol = "start", callProbMethod = "CGHCall", callProbabilities = false, chrColumn = "Chr", filterNA = true, lowerLimit = 0.0, minWidth = 2, nSegFit = 3000, outputAllSegs = false, plotChromosomes = "0", plotEverySample = false, undoSD = 3, undoSplits = "sdundo", upperLimit = 0.0 )

Inputs

Name	Type	Mandatory	Description
caseChan	CSV	Mandatory	CSV file containing the normalized probe intensities for the case channel. First row should have the probenames and samples should be columnwise.
geneAnnotation	AnnotationTable	Mandatory	Probewise annotations as produced by the AgilenReader component. Only the probes present in the casechannel csv can be included. Use CSVFilter if the probes do not otherwise match.

Outputs

Name	Type	Description
report	Latex	Latex report for the analysis.
segments	CSV	Segmented probewise copy number change values.
tholds	CSV	Thresholds that were used to define a segment aberrated.
lossProbs	CSV	Probewise probability of a loss having occured in the segment to which the probe was assigned.
gainProbs	CSV	Probewise probability of a gain having occured in the segment to which the probe was assigned.
normProbs	CSV	Probewise probability of no copy number change in the segment to which the probe was assigned.
rawSegments	CSV	Probewise segment means.
frequency	CSV	Probewise copy-number alteration frequencies.

Parameters

Name	Type	Default	Description
CGHCallMaxnumseg	int	100	Only used if callProbMethod=CGHCall. Maximum number of segments on a sample to be used for fitting the probabilty model.
CGHCallPrior	string	"auto"	Only used if callProbMethod=CGHCall. Set the method to determine prior probabilities to CGHCall algorithm. Must be one of "auto", "all", "not all".
CGHCallRobustsig	boolean	true	Only used if callProbMethod=CGHCall. Setting this to true enforces a lower bound on the normal segments.
alpha	float	0.01	P-value of CBS to accept a break point.
bpEndCol	string	"end"	Column name for chromosome end basepair in geneAnnotation.
bpStartCol	string	"start"	Column name for chromosome start basepair in geneAnnotation.
callProbMethod	string	"CGHCall"	Method with which to call CNA segment probabilities. Must be either CGHCall or FastCall. FastCall is significantly (30000 times) faster while CGHCall is more accurate.
callProbabilities	boolean	false	Enabling this will make the component estimate probabilties for CNA segments via the CGHcall package. Only guaranteed to work when multiple chromosomes are analyzed simultaneously.
chrColumn	string	"Chr"	Column name for chromosome identifiers in geneAnnotation.
filterNA	boolean	true	Filter out probes with NA values. This is always true when 'callProbabilities' is true.
lowerLimit	float	0.0	Use this as the intensity limit for calling a segment lossed. Default value computes the mean of the sample set and estimates the lower threshold for CNA call to be two standard deviations from it.
minWidth	int	2	Minimum number of probes for CBS to define a segment. CBS only allows widths between 2 and 5.
nSegFit	int	3000	Maximum number of segments used for fitting the mixture model in CGHcall probability calculations. Disabled if callProbabilities=false. Decreasing this lowers accuracy but can speed computation significantly.
outputAllSegs	boolean	false	Output all segmentation results even if non-significant or non-aberrated.
plotChromosomes	string	"0"	Defines the plots that the user wants to output as a comma separated list. The whole genome is plotted by default for each sample. Inputting a value different than 0 will generate additional plots for these specific chromosomes from each sample.
plotEverySample	boolean	false	Enabling this will make the component print each sample separately.
undoSD	int	3	Only used if 'undoSplits=sdundo'. Defines how many SDs two adjacent segments can be apart before they are combined.
undoSplits	string	"sdundo"	A character string specifying how change-points are to be undone, if at all. Undoing change-points decreases the sensitivity of segmentation. Choices are "none","prune", which uses a sum of squares criterion, and "sdundo" (default), which undoes splits that are not at least this many SDs apart. SD by default is 3.
upperLimit	float	0.0	Use this as the intensity limit for calling a segment gained. Default value computes the mean of the sample set and estimates the upper threshold for CNA call to be two standard deviations from it.

Test cases

Test case	Parameters▼	IN caseChan	IN geneAnnotation	OUT report	OUT segments	OUT tholds	OUT lossProbs	OUT gainProbs	OUT normProbs	OUT rawSegments	OUT frequency
case1	properties	caseChan	geneAnnotation	report	segments	tholds	lossProbs	gainProbs	normProbs	rawSegments	(missing)
	chrColumn=Chromosome, bpStartCol=startBp, bpEndCol=endBp, filterNA=true
case2	properties	caseChan	geneAnnotation	report	segments	tholds	lossProbs	gainProbs	normProbs	(missing)	(missing)
	plotChromosomes = 6,7, plotEverySample = true, callProbabilities = true, nSegFit = 500, CGHCallPrior = all, lowerLimit = -0.650366166469952, upperLimit = 0.587166166469953, filterNA = false
case3	properties	caseChan	geneAnnotation	report	segments	tholds	(missing)	(missing)	(missing)	(missing)	(missing)
	upperLimit=0.583291228317388, lowerLimit=-0.646491228317388, callProbabilities=false
case4_InvalidParam	properties	caseChan	geneAnnotation	(expecting failure)	(expecting failure)	(expecting failure)	(expecting failure)	(expecting failure)	(expecting failure)	(expecting failure)	(expecting failure)
	minWidth=6, undoSplits=none
case5_invalidColumnNames	properties	caseChan	geneAnnotation	(expecting failure)	(expecting failure)	(expecting failure)	(expecting failure)	(expecting failure)	(expecting failure)	(expecting failure)	(expecting failure)
	undoSplits=prune
case6_allSegs	properties	caseChan	geneAnnotation	(missing)	segments	(missing)	(missing)	(missing)	(missing)	(missing)	(missing)
	upperLimit=1, lowerLimit=-1, callProbabilities=false, outputAllSegs=true

Test network

Display TestNetwork

import org.anduril.runtime._
import anduril.microarray._
import anduril.builtin._
object testnetworks_ACGHSegment { 
 
val in1 = INPUT(path="../../components/ACGHnorm/testcases/case3/input/caseIn")
val in2 = INPUT(path="../../components/ACGHnorm/testcases/case3/input/control")
val inA = INPUT(path="../../components/ACGHnorm/testcases/case3/input/probeAnnotation")
 
val normal = ACGHnorm(caseIn = in1,
                  control     = in2,
                  method      = "loess")
 
val segmented = ACGHsegment(caseChan       = normal.casechannel,
                        geneAnnotation = inA,
                        outputAllSegs  = true,
                        chrColumn  = "chromosome",
                        bpStartCol = "startLocus",
                        bpEndCol   = "endLocus",
                        minWidth   = 5,
                        lowerLimit = -0.2,
                        upperLimit = 0.2)
 
OUTPUT(segmented.segments)
}

Generated 2019-02-08 07:42:08 by Anduril 2.0.0