Up: Component summary Component

CNGEIntegrator

Performs one of two possible nonparametric tests for the detection of copy number induced differential gene expression. The tests incorporate the uncertainty of the calling of genomic aberrations: weighted versions of Cramer-von-Mises and Mann-Whitney test statistics are used. An efficient permutation re-sampling procedure is used for p-value calculation. The test statistics may be "shrunken" to borrow information across neighboring genes that share the same copy number signature.

The intCNGEan and CGHcall R packages can be be found here and here.

Version 1.1.0
Bundle microarray
Categories Integration Copy Number Analysis
Authors Riku Louhimo (Riku.Louhimo@Helsinki.FI)
Issue tracker View/Report issues
Requires intCNGEan (R-package) ; impute (R-bioconductor) ; CGHcall (R-bioconductor) ; DNAcopy (R-bioconductor)
Source files component.xml CNGEIntegrator.r
Usage Example with default values

Inputs

Name Type Mandatory Description
CNAProbLoss CSV Mandatory Matrix of probabilities for probewise copy number losses as output by ACGHsegment component.
CNAProbGain CSV Mandatory Matrix of probabilities for probewise copy number gains as output by ACGHsegment component.
CNAProbNorm CSV Mandatory Matrix of probabilities for probewise copy number alterations as output by ACGHsegment component.
exprMatrix CSV Mandatory Expression matrix.
exprAnnotation AnnotationTable Mandatory Probe annotations for the expression matrix. Must contain columns 'Chr' and 'Start'.

Outputs

Name Type Description
integrationResult CSV Results of the integration. If analysisType="univariate", the component returns a CSV with the following columns: the first three columns contain the annotation information; the gene number of the original (pre-tuned, matched) data set; indicator of test performed, either "1" (loss vs. no-loss) or "2" (no-gain vs. gain); the estimated marginal call probability distribution (two columns); the estimated copy number induced gene expression change; the coefficient of determination; the raw p-value; the BH-corrected p-value. If analysis.type="regional", the component returns a CSV with the following columns: the first three columns contain the annotation information; the gene number of the original (pre-tuned, matched) data set; indicator of test performed, either "1" (loss vs. no-loss) or "2" (no-gain vs. gain); the estimated marginal call probability distribution (two columns); the estimated copy number induced gene expression change; the coefficient of determination; the region number within chromosome; gene number of first and last gene within the region (two columns); shrinkage parameter; the raw p-value; the BH-corrected p-value.

Parameters

Name Type Default Description
FDRLimit float 0.05 False-discover rate corrected p-value threshold.
analysisType string "univariate" Determine whether to shrink test statistic or not. Defaults to non-shrinkage.
matchMethod string "overlap" Matching occurs on the basis of genomic locations. In case matchMethod="distance", the midpoint of copy number and expression probes are calculated and for each gene on the expression array the closest feature of the copy number platform is selected. If matchMethod="overlap", each gene in the expression array is matched to the feature from the copy number platform with the maximum percentage of overlap. If the maximum percentage of overlap equals zero, the gene is not included in the matched objects. If matchMethod="overlapplus", the features are first matched by their percentage of overlap (as with the matchMethod="overlap"-option). For all non-matched expression probes, the closest two copy number features (one down- and one upstream) are determined. If the copy number signature of these two copy number probes is identical, intrapolation seems reasonable, and the expression probe is matched to the closest of these two CN probes. Hence, matchMethod="overlapplus" makes use of the copy number data, consequently, matching may be different for different data sets.
nGenesTune int 50 Number of genes used in the tuning step.
nperm int (no default) Number of permutations in the integration permutation test.
testStatistic string "wmw" Test statistic to be used. Either "wcvm" for weighted Cramer-von-Mises statistic or "wmw" for weighted Mann-Whitney test statistic.

Test cases

Test case Parameters IN
CNAProbLoss
IN
CNAProbGain
IN
CNAProbNorm
IN
exprMatrix
IN
exprAnnotation
OUT
integrationResult
case1 properties CNAProbLoss CNAProbGain CNAProbNorm exprMatrix exprAnnotation integrationResult

FDRLimit=0.5,
analysisType=regional,
nperm=10

case2 properties CNAProbLoss CNAProbGain CNAProbNorm exprMatrix exprAnnotation integrationResult

matchMethod=distance,
nperm=10


Generated 2018-12-18 07:42:17 by Anduril 2.0.0