An R package for inferring the subclonal architecture of tumors

Installation instructions:

As of mid-2022, the NORMT3 package, which is a dependency of sciclone/bmm, has been removed from CRAN. It can be installed manually by doing something like:

$ wget https://cran.r-project.org/src/contrib/Archive/NORMT3/NORMT3_1.0.4.tar.gz
$ R CMD install NORMT3_1.0.4.tar.gz

Then proceed with the below instructions:

Both the 'sciClone' package and it's 'bmm' dependency can be installed by doing the following:

#install IRanges from bioconductor
source("http://bioconductor.org/biocLite.R")
biocLite("IRanges")
#install devtools if you don't have it already
install.packages("devtools")
library(devtools)
install_github("genome/bmm")
install_github("genome/sciClone")

If you prefer to build the package by hand, follow these steps:

• Make sure that you have the dependencies from the CRAN and BioConductor repos: IRanges, rgl, RColorBrewer, ggplot2, grid, plotrix, methods, NORMT3, MKmisc, TeachingDemos

• install the bmm package from https://github.com/genome/bmm

• Download and build from source:

    git clone [email protected]:genome/sciclone.git
    R CMD build sciclone
    R CMD INSTALL sciClone_1.1.0.tar.gz
  

Usage

library(sciClone)

#read in vaf data from three related tumors
#format is 5 column, tab delimited: 
#chr, pos, ref_reads, var_reads, vaf

v1 = read.table("data/vafs.tumor1.dat",header=T);
v2 = read.table("data/vafs.tumor2.dat",header=T);
v3 = read.table("data/vafs.tumor3.dat",header=T);

#read in regions to exclude (commonly LOH)
#format is 3-col bed
regions = read.table("data/exclude.loh")

#read in segmented copy number data
#4 columns - chr, start, stop, segment_mean   
cn1 = read.table("data/copy_number_tum1")
cn2 = read.table("data/copy_number_tum2")
cn3 = read.table("data/copy_number_tum3")

#set sample names
names = c("Sample1","Sample2","Sample3")


#Examples:
#------------------------------------
#1d clustering on just one sample
sc = sciClone(vafs=v1,
         copyNumberCalls=cn1,
         sampleNames=names[1],
         regionsToExclude=reg1)
#create output
writeClusterTable(sc, "results/clusters1")
sc.plot1d(sc,"results/clusters1.1d.pdf")

#------------------------------------
#2d clustering using two samples:
sc = sciClone(vafs=list(v1,v2),
              copyNumberCalls=list(cn1,cn2),
              sampleNames=names[1:2],
               regionsToExclude=regions)
#create output
writeClusterTable(sc, "results/clusters2")
sc.plot1d(sc,"results/clusters2.1d.pdf")
sc.plot2d(sc,"results/clusters2.2d.pdf")


#------------------------------------
#3d clustering using three samples:
sc = sciClone(vafs=list(v1,v2,v3),
              copyNumberCalls=list(cn1,cn2,cn3),
              sampleNames=names[1:3],
               regionsToExclude=regions)
#create output
writeClusterTable(sc, "results/clusters2")
sc.plot1d(sc,"results/clusters2.1d.pdf")
sc.plot2d(sc,"results/clusters2.2d.pdf")
sc.plot3d(sc, sc@sampleNames, size=700, outputFile="results/clusters3.3d.gif")

#This pattern generalizes up to N samples, except for plotting, which caps out at 3d for obvious reasons.

Visualization

single-tumor plot

2d comparison plot

3d comparison plot

Notes

• Requires host system to have imagemagick installed before it can produce animated gif output of 3d plots.

• Input formats described in more detail in the R documentation (see ?sciClone)

• Many questions regarding sciClone usage have been asked and answered on Biostar: https://www.biostars.org/t/sciclone/

Accessory Scripts and Data

The sciClone-meta repo contains all data and scripts used to create the figures in the manuscript. It also contains a small suite of tests that demonstrate the capabilities of sciClone and verify that it is installed correctly.

Reference

Manuscript published at PLoS Computational Biology (doi:10.1371/journal.pcbi.1003665)

SciClone: Inferring clonal architecture and tracking the spatial and temporal patterns of tumor evolution

Christopher A. Miller^1*, Brian S. White^2*, Nathan D. Dees¹, John S. Welch^2,3, Malachi Griffith¹, Obi Griffith¹, Ravi Vij^2,3, Michael H. Tomasson^2,3, Timothy A. Graubert^2,3, Matthew J. Walter^2,3, William Schierding¹, Timothy J. Ley^1,2,3, John F. DiPersio^2,3, Elaine R. Mardis^1,3,4, Richard K. Wilson^1,3,4, and Li Ding^1,2,3,4

¹The Genome Institute

²Department of Medicine

³Siteman Cancer Center

⁴Department of Genetics Washington University, St. Louis, MO 63110, USA

^* These authors contributed equally to this work