Generating realistic data with known truth using the jointseg package

M. Pierre-Jean, G. Rigaill, P. Neuvial

2019-01-11

This vignette describes how to use the jointseg package to partition bivariate DNA copy number signals from SNP array data into segments of constant parent-specific copy number. We demonstrate the use of the PSSeg function of this package for applying two different strategies. Both strategies consist in first identifying a list of candidate change points through a fast (greedy) segmentation method, and then to prune this list is using dynamic programming [1]. The segmentation method presented here is Recursive Binary Segmentation (RBS, [2]). We refer to [3] for a more comprehensive performance assessment of this method and other segmentation methods. segmentation, change point model, binary segmentation, dynamic programming, DNA copy number, parent-specific copy number.

Please see Appendix for citing jointseg.

HERE

This vignette illustrates how the jointseg package may be used to generate a variety of copy-number profiles from the same biological ``truth’’. Such profiles have been used to compare the performance of segmentation methods in [3].

Citing jointseg

citation("jointseg")
## 
## To cite package 'jointseg' in publications, please use the
## following references:
## 
##   Morgane Pierre-Jean, Guillem Rigaill and Pierre Neuvial ().
##   jointseg: Joint segmentation of multivariate (copy number)
##   signals.R package version 1.0.2.
## 
##   Morgane Pierre-Jean, Guillem Rigaill and Pierre Neuvial.
##   Performance evaluation of DNA copy number segmentation methods.
##   Briefings in Bioinformatics (2015) 16 (4): 600-615.
## 
## To see these entries in BibTeX format, use 'print(<citation>,
## bibtex=TRUE)', 'toBibtex(.)', or set
## 'options(citation.bibtex.max=999)'.

Setup

The parameters are defined as follows:

n <- 1e4                                 ## signal length
bkp <- c(2334, 6121)                     ## breakpoint positions
regions <- c("(1,1)", "(1,2)", "(0,2)")  ## copy number regions
ylims <- cbind(c(0, 5), c(-0.1, 1.1))
colG <- rep("#88888855", n)
hetCol <- "#00000088"

For convenience we define a custom plot function for this vignette:

plotFUN <- function(dataSet, tumorFraction) {
    regDat <- acnr::loadCnRegionData(dataSet=dataSet, tumorFraction=tumorFraction)
    sim <- getCopyNumberDataByResampling(n, bkp=bkp,
                                         regions=regions, regData=regDat)
    dat <- sim$profile
    wHet <- which(dat$genotype==1/2)
    colGG <- colG
    colGG[wHet] <- hetCol
    plotSeg(dat, sim$bkp, col=colGG)
}

Affymetrix data

ds <- "GSE29172"
pct <- 1
plotFUN(ds, pct)
Data set GSE29172 : 1 % tumor cells

Data set GSE29172 : 1 % tumor cells

plotFUN(ds, pct)
Data set GSE29172 : 1 % tumor cells (another resampling)

Data set GSE29172 : 1 % tumor cells (another resampling)

pct <- 0.7
plotFUN(ds, pct)
Data set GSE29172 : 0.7 % tumor cells

Data set GSE29172 : 0.7 % tumor cells

pct <- 0.5
plotFUN(ds, pct)
Data set GSE29172 : 0.5 % tumor cells

Data set GSE29172 : 0.5 % tumor cells

Illumina data

ds <- "GSE11976"

Session information

sessionInfo()
## R version 3.5.1 (2018-07-02)
## Platform: x86_64-apple-darwin15.6.0 (64-bit)
## Running under: OS X El Capitan 10.11.6
## 
## Matrix products: default
## BLAS: /Library/Frameworks/R.framework/Versions/3.5/Resources/lib/libRblas.0.dylib
## LAPACK: /Library/Frameworks/R.framework/Versions/3.5/Resources/lib/libRlapack.dylib
## 
## locale:
## [1] C/fr_FR.UTF-8/fr_FR.UTF-8/C/fr_FR.UTF-8/fr_FR.UTF-8
## 
## attached base packages:
## [1] stats     graphics  grDevices utils     datasets  methods   base     
## 
## other attached packages:
## [1] jointseg_1.0.2 knitr_1.20    
## 
## loaded via a namespace (and not attached):
##  [1] Rcpp_1.0.0         matrixStats_0.54.0 digest_0.6.18     
##  [4] rprojroot_1.3-2    acnr_1.0.0         backports_1.1.2   
##  [7] magrittr_1.5       evaluate_0.12      highr_0.7         
## [10] stringi_1.2.4      rmarkdown_1.10     tools_3.5.1       
## [13] stringr_1.3.1      yaml_2.2.0         compiler_3.5.1    
## [16] htmltools_0.3.6    DNAcopy_1.54.0

References

[1] Bellman, Richard. 1961. “On the Approximation of Curves by Line Segments Using Dynamic Programming.” Communications of the ACM 4 (6). ACM: 284.

[2] Gey, Servane, et al. 2008. “Using CART to Detect Multiple Change Points in the Mean for Large Sample.” https://hal.archives-ouvertes.fr/hal-00327146.

[3] Pierre-Jean, Morgane, et al. 2015. “Performance Evaluation of DNA Copy Number Segmentation Methods.” Briefings in Bioinformatics, no. 4: 600-615.