Repository files navigation

James Dalgleish

January 24, 2021Certificate updated today! Message me if any have trouble accessing the server.

• Installation
• App
• Package
• Video Tutorial
• Additional Examples

CNVScope is a visualization toolkit for seeing copy number data in arelationship fashion, both utilizing tools that find regions of interest(much like domains in Hi-C terminology), visualize the relationships invibrant color (blue for negative association, red for positive), andintegrates several different relevant types of data (RNASeq, sampleinformation, Cancer Gene Census data, and structural variants).Structural variants like rearrangements can be thought of asinteractions between genomic regions, which is why it makes sense tobuild GenomicInteractions objects and from this, interaction matricesthat can be easily visualized.

The package is intended for users with experience in the R programming language.Non-R users should use the public site:http://cnvscope.nci.nih.gov

The program should install using the following code:

install.packages("CNVScope")

To install the development version (may have more features), please use:

remotes::install_github("jamesdalg/CNVScope")CNVScope::runCNVScopeLocal()

For those having difficulty, we provide more detailed instructions below:

Installation works best if base R and R tools are up to date.

1. install R directly from the Comprehensive R Archive Network (CRAN):https://cran.r-project.org/index.htmlWindows:https://cran.r-project.org/bin/windows/base/Macintosh:https://cran.r-project.org/bin/macosx/
2. Install RStudio. If running macintosh, install Xcode when prompted.
3. Install the appropriate compilation tools for your operating system:a. For windows, please download & install the latest version of RTools:https://cran.r-project.org/bin/windows/Rtools/b. For macintosh, be sure to install the latest version of gfortran and clang (make sure these install AFTER Xcode, so that it overwrites the existing clang installation):https://cran.r-project.org/bin/macosx/tools/
4. Install Bioconductor Dependencies, set allowed repositories, and install the package:Run the following on the R command line:

install.packages("BiocManager")BiocManager::install(c("DNAcopy","GenomeInfoDb","BSgenome.Hsapiens.UCSC.hg19"))setRepositories(ind=c(1:2))install.packages("CNVScope")

The package is now installed and should contain the vignettes, functions, and help files.

5. To install the app data (several GB), use the following lines:

install.packages("remotes")CNVScope::runCNVScopeLocal()

Please wait for the package to finish downloading. It is a large package that contains allof the Neuroblastoma Data and may take some time.In the meantime, feel free to browse our public server:https://cnvscope.nci.nih.gov

The app includes several components and starts with a customizablecontrol panel. As an example, we’ll search for TERT and TP53.

Doing it this way creates a zoomed in plot of the copy numberrelationships between the chromosomes containing TERT (chromosome 5) andTP53 (chromosome 17), zoomed into the specific region for thoseparticular genes (as denoted in the search box). For many regions, wecan rely on the tooltip for genes, but this region is particularly genedense and it’s best to simply click on the region of interest. The rowand column genes are searchable here and will aid us to understand thatwe’re in the right place.

Clicking on the cancer gene census tab will lead us to the genes thatare found within the cancer gene census, (a curated list of cancergenes)[https://cancer.sanger.ac.uk/census]. Corresponding infosuggests what cancers associated with the gene, mechanism of action (ifknown), and tissue types.

The sample level information will lead us to a series of plots(histogram, scatter, a colored linear regression plot) that will get atthe underlying data behind the negative log p-value, and help us detectoutliers, look at clusterings of samples, and see if the regressionp-value is driven by a variety of points from different cancer typesrather than just an outlier or a group of outliers. All the plots arezoomable and a slider for opacity has been provided for the histogram tomore clearly see distributions that might tightly overlap.

Expression data is provided for neuroblastoma, sorting the most variablegenes to the top for the specific set of chromosomal regions for theclicked point.

Finally, a whole genome view is provided to give the viewer a sense ofthe complexity and domains of the neuroblastoma interactome and toencourage further exploration of subpoints. The slider allows the userto select the saturation limit and thereby see domains that otherwisewould be invisible due to the high values on the diagonal.

The package focuses on methods of analyzing these matrices andconstructing components for app use. See the following vignettes formore information (also available on command line):

Creating the Input matrix from publicdata

LinearRegression/Postprocess

For a complete video tutorial (16 minutes in length),clickhere.

Feel free to look at the additional examples vignette. We have severalGDC datasets that demonstrate that our package is a versatile toolkitthat is useful for NBL, as well as SKCM (Skin Cancer), BLCA (bladdercancer), AML (Acute Myleoid Leukemia), and PRAD (Prostate Cancer). Wealso demonstrate our copy number relationship weighting function as wellas contour and 3D representations. The final example, of BLCA chr17, isshown below.Click here to download an interactive HTMLversion.