XenofilteR: computational deconvolution of mouse and human reads in tumor xenograft sequence data
Human tumour samples or cancer cell lines, transplanted into mice are widely used as a model to study cancer. However, genomic analysis of tumour material derived from these xenografts is challenging. The sequence data not only contains reads that originate from the graft (human tumour or cell line) but also reads from host (mouse). We developed XenofilteR, an R-package for filtering host from graft reads in next generation sequence data. XenofilteR is a novel method that utilizes the edit distance of each read for classification. XenofilteR outperforms existing methods as validated on artificially mixed mouse/human samples and sets of patient derived xenograft samples.
The paper that accompanies XenofilteR has been published in BMC Bioinformatics: [XenoFilteR paper] (https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-018-2353-5)
Kluin, R. J. C. et al. XenofilteR: computational deconvolution of mouse and human reads in tumor xenograft sequence data. BMC Bioinformatics 19, 366 (2018).
Installation of XenofilteR should be performed as follows:
> source("http://bioconductor.org/biocLite.R") > biocLite(c("Rsamtools", "GenomicAlignments", "BiocParallel", "futile.logger"))
As the last step in the installation process, the latest XenofilteR package can be downloaded from the XenoFilteR releases webpage and installed using the following command:
$ R CMD INSTALL XenofilteR*.tar.gz
Now you are all set to start your analysis.
Load the XenofilteR package in R using:
XenofilteR fully supports paralel computing and is implemented in such a way that every sample is processed on a single core. XenofilteR uses the BiocParallel for parallel computing. This package requires the user to specify which parallel environment is used by creating an instance of BiocParallelParam. Here, we use the 'snow' environment and specify a SnowParam. The number of workers represents the number of CPUs used for the analysis and thereby the number of samples analyses simultaneously. Hence, it is wise to keep the number of CPUs low when analysing large samples as memory usage might be high.
> bp.param <- SnowParam(workers = 1, type = "SOCK")
XenofilteR requires a dataframe or matrix, named 'sample.list', with in the first column the bam file names as mapped to the graft reference. The second column contains the file names and paths to the bam files as mapped to the host reference. Each row in 'sample.list' represents a single sequence run or sample. An optional list may be provided with alternative names for output files, 'output.names '. Especially for RNAseq samples aligned with for example Tophat this may be convenient since all .bam files are named identical. The XenofilteR package and data are run in the following way:
> XenofilteR(sample.list, destination.folder = "./", bp.param = bp.param, output.names)
XenofilteR runs without issues on RNAseq mapped with STAR. However, the 'NM'-tag in the BAM file is not generated by STAR with default settings. This tag is mandatory for XenofilteR to run. Please add the following to your STAR command when mapping to the human as well as the mouse reference genome:
Additional explanation about the 'NM'-tag and optional output from STRA can be found here: STAR manual (https://github.com/alexdobin/STAR/blob/master/doc/STARmanual.pdf) (section 9.11)
We have tried to make the XenofilteR code readable and its use as easy as possible. If any questions arise regarding the package, or if you want to report any bugs, please do not hesitate and contact:
Oscar and Roel both work at the NKI, Oscar as a postdoc in the lab of Prof. Daniel Peeper, Roel as bioinformatician in the Genome Core Facility.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.