Basal-like Breast Cancer DNA copy number losses
identify genes involved in genomic instability,
response to therapy, and patient survival

Victor J. Weigman^1,2,3,*, Hann-Hsiang Chao^4,*, Andrey A. Shabalin^5,8, Xiaping He^2,4, Joel S. Parker^2,4, Silje Nordgard⁹, Tatyana Grushko¹⁰, Dezheng Huo¹⁰, Chika Nwachukwu¹⁰, Andrew Nobel^5,8, Vessela N. Kristensen^9,11,12 Anne-Lisa Børresen-Dale^9,12, Olufunmilayo I. Olopade¹⁰, Charles M. Perou^2,4,6,7

SupWald Identification of dna-copy CHanges (SWITCHdna)

Supplemental Website

¹ Program in Bioinformatics and Computational Biology,
² Lineberger Comprehensive Cancer Center,
³ Department of Biology,
⁴ Department of Genetics,
⁵ Department of Statistics and Operations Research,
⁶ Carolina Center for Genome Sciences,
⁷ Department of Pathology and Laboratory Medicine,
⁸ Department of Biostatistics

University of North Carolina at Chapel Hill, Chapel Hill, NC 27599

⁹ Department of Genetics, Institute for Cancer Research, Oslo University Hospital, Radiumhospitalet, Norway
¹⁰ Center for Clinical Cancer Genetics and Global Health, University of Chicago Medical Center, Chicago, IL 60615

¹¹ Department of Clinical Molecular Biology (EpiGen), Akershus University Hospital, University of Oslo, Oslo, Norway

¹² Institute for Clinical Medicince, Faculty of Medicine, University of Oslo, Oslo, Norway

*=these authors made equal contributions

# = Corresponding author

In the paper we introduce a new method, SWITCHdna, to identify regions of Copy Number Aberrations (CNAs) based on intensity data generated from copy number platforms. The method is a modification of change point detection method used by Lickwar et al. (2009). The SWITCHdna method performs identification of transition points followed by significance testing of defined segments. The package also includes functions for visual representation of the findings. The method is described in the paper and the instructions are provided in the manual. In SWITCHdna, we perform estimation of transition points sequentially. First, we test for the presence of at least one transition point (versus none) across the entire chromosome. To conduct the test we calculate F-statistic for the difference of means for every candidate (possible) transition point such that the size of the region on either side of the breakpoint was larger than ±. If the maximum observed F-statistic is below a user-specified threshold, in other words the test shows no significant deviation from the null model (or no change points), we accept K=0. Otherwise, we say that the transition happened at the location of the maximum F-statistic. The testing procedure is then applied separately to the regions to the left and to the right of the transition point. Following detection of the transition points, for each segment we record its average value, number of observations in the region, and the corresponding z-score. We filter out segments with a z-score less than 3 and a average intensity measurement of 0.09 or less. The code was tested on both Windows and Linux systems on both 32 and 64 bit versions of R and on Revolution R. Questions about the package should be sent to weigman-AT-gmail.com.

Download: SWITTCHdna R code. Requires R programming languge. SWITTCHdna Manual. Adobe .pdf format, may require Adobe Reader. Sample input data. Includes sample aCGH data, normalized and Log-transformed, and supplementary files. The landscape plot presented below can be generated via analysis of the sample data using the code in readme.txt file.

Visualisation of the findings.

The SWITCHdna package includes functions for visual representation of the findings. They generate plots which show the frequencies of CNAs within the groups across the genome. For example, analysis of the provided sample data produces the following landscape plot:

(click the picture for bigger version)