extractIBDsegments: Extract IBD segments from a 'fabia' result
In hapFabia: hapFabia: Identification of very short segments of identity by descent (IBD) characterized by rare variants in large sequencing data

Description Usage Arguments Details Value Author(s) References See Also Examples

extractIBDsegments: R implementation of extractIBDsegments.

IBD segments are identified in FABIA Factorization objects. First accumulations of correlated SNVs are found. Then IBD segments in these accumulations are disentangled. Finally IBD segments are pruned off spurious correlated SNVs.

## S4 method for signature 
## 'Factorization,
##   list,
##   data.frame,
##   character,
##   matrix,
##   numeric,
##   numeric,
##   numeric,
##   numeric,
##   numeric,
##   numeric,
##   numeric,
##   numeric'
extractIBDsegments(res,sPF,annot=NULL,chrom="",labelsA=NULL,ps=0.9,psZ=0.8,inteA=500,thresA=11,mintagSNVs=8,off=0,procMinIndivids=0.1,thresPrune=1e-3)

`res`	result of `fabia` given as `Factorization` object.
`sPF`	genotype data obtained by `fabia` procedure `samplesPerFeature`; it gives for each SNV the individuals/chromosomes that possess the minor allele.
`annot`	annotation for the tagSNVs as an object of the class `data.frame`; if it is `NULL` then a dummy annotation is generated.
`chrom`	the chromosome the genotyping data stems from.
`labelsA`	labels for the individuals; if it is `NULL` then dummy labels by enumerating individuals are generated.
`ps`	quantile above which the L values are considered for IBD segment extraction.
`psZ`	quantile above which the largest Z values are considered for IBD segment extraction.
`inteA`	number of SNVs in a histogram bin which correspond to the desired IBD segment length.
`thresA`	threshold for histogram counts above which SNVs are viewed to be locally accumulated in a histogram bin.
`mintagSNVs`	threshold for minimal tagSNV overlap of intervals in a IBD segment.
`off`	offset of the histogram.
`procMinIndivids`	percent of cluster individuals that must have the minor allele to consider an SNV as IBD segment tagSNV.
`thresPrune`	threshold on the probability of having minimal distance to neighboring tagSNVs; used to prune off SNVs at the border of IBD segments.

The threshold thresA for counts in a bin, which indicates SNV accumulations, is computed and provided by hapFabia when calling this method. Distance probabilities for pruning are based on an exponential distribution with the median distance between tagCNVs as parameter (one over the rate). Thus, the counts are assumed to be Poisson distributed. At the IBD segment border, SNVs that have a large distance to the closest tagSNV are pruned off. thresPrune gives the pruning threshold via a p-value for observing this distance or a larger based on the exponential distribution.

Implementation in R.

An instance of the class IBDsegmentList containing the extracted IBD segments.

Sepp Hochreiter

S. Hochreiter et al., ‘FABIA: Factor Analysis for Bicluster Acquisition’, Bioinformatics 26(12):1520-1527, 2010.

IBDsegment-class, IBDsegmentList-class, analyzeIBDsegments, compareIBDsegmentLists, extractIBDsegments, findDenseRegions, hapFabia, hapFabiaVersion, hapRes, chr1ASW1000G, IBDsegmentList2excel, identifyDuplicates, iterateIntervals, makePipelineFile, matrixPlot, mergeIBDsegmentLists, mergedIBDsegmentList, plotIBDsegment, res, setAnnotation, setStatistics, sim, simu, simulateIBDsegmentsFabia, simulateIBDsegments, split_sparse_matrix, toolsFactorizationClass, vcftoFABIA

data(hapRes)
res <- hapRes$res
sPF <- hapRes$sPF
annot <- hapRes$annot
nnL <- length(Z(res)[1,])
labelsA <- cbind(as.character(1:nnL),
   as.character(1:nnL),as.character(1:nnL),
   as.character(1:nnL))
resIBDsegmentList <- extractIBDsegments(res=res,
   sPF=sPF,annot=annot,chrom="1",labelsA=labelsA,
   ps=0.9,psZ=0.8,inteA=50,thresA=6,mintagSNVs=6,
   off=0,procMinIndivids=0.1,thresPrune=1e-3)

summary(resIBDsegmentList)

print("Position of the first IBD segment:")
print(IBDsegmentPos(resIBDsegmentList[[1]]))

print("Length of the first IBD segment:")
print(IBDsegmentLength(resIBDsegmentList[[1]]))