BreakDown
| Overview | |
| Description | BreakDown genotyping SVs in heterogeneous tumor sample |
| Development Information | |
| Language | Perl |
| Current version | 1.0 |
| License | GPL v3 |
| Status | Active |
| References | |
| Citation | Fan, Zhou, Chong, Nakhleh, and Chen, Towards accurate characterization of clonal heterogeneity based on structural variation, BMC Bioinformatics (2014) 15:299. https://doi.org/10.1186/1471-2105-15-299 |
| Help and Support | |
| Contact | Ken Chen |
BreakDown is a perl package that estimates the variant allele fraction (VAF) and infers genotypes of structural variants from next generation paired-end sequencing reads. There are two steps in BreakDown. First, bam2cfg.pl is used to generate a configure file of the sequence data. Next, BreakDown.pl uses this configure file and the putative SV files to generate a VAF/genotype summary file.
Exciting development in deep digital sequencing technology has started to reveal fascinating clonal structure in tumor samples. Accurately depicting the architecture of such intra-tumor clonality and understanding their clinical relevance will require analytic methods that can sufficiently and accurately utilize both the sequence coverage and the physical coverage in the sequencing data. We develop BreakDown, a novel computational method that aims to utilize all types of structural variants (SVs) in clonal heterogeneity inference. Previous work in this field has involved straightforward counting of reads that span single nucleotide variants or small indels but has largely ignore SVs that are abundant in genomically instable tumor samples. Some recent methods such as THetA have investigated large copy number alterations but not other types of SVs such as translocations, inversions, etc. They also have not utilized the most accurate evidences in the alignment data: reads or read pairs that span or overlap the SV breakpoints.
Our study is novel in the following aspects:
1) it is the first that presents a unified analytical model that can probabilistically integrate evidences from all types of reads (concordantly mapped, discordantly mapped, or partially mapped) and is therefore more accurate than published approaches;
2) it supports more SVs (including balanced SVs) in wider size ranges (from hundred bp to Megabases) than published methods;
3) it has been implemented in a practical bioinformatics tool BreakDown that has been validated using real heterogeneous tumor data with previously known clonal structure;
4) it follows the working standard in both cancer genomics (such as TCGA) and the population genomics communities (such as 1000 genomes) and compute quantities (such as genotype likelihood and variant allele fraction) that can be conveniently utilized by other tools.
BreakDown-1.0 (358 KB)