Ssion The PREGO algorithm presented here combines copy number and adjacencySsion The PREGO algorithm presented

Ssion The PREGO algorithm presented here combines copy number and adjacency
Ssion The PREGO algorithm presented here combines copy number and adjacency information from paired-end sequencing data to infer cancer genome organization. However, the algorithm does not consider all the issues involved in real cancer sequencing data. In particular, we assume that structural variants can be identified by mapping of discordant paired reads, but this is difficult for structural variants in repetitive regions of the human genome [15,17]. Thus, there may be missing or incorrect adjacencies in the data. Similarly, estimates of read depth are difficult to obtain in repetitive regions [21]. While some of these issues may be addressed computationally, the more difficult cases will require longer reads and/or longer fragments for paired reads. Beyond the issues with data quality are limitations on the inferred organization. While we derive multiplicities on the edges using adjacency and copy number data, PubMed ID:https://www.ncbi.nlm.nih.gov/pubmed/28667899 we do not resolve the resulting paths through the intervaladjacency graph, except in simple cases. In many datasets, there will be many such paths and therefore many reconstructions of the cancer genome that are consistent with the data. Even the solution for the estimated edge multiplicities may not be unique. Resolving such longer paths requires additional information about connections between consecutive adjacencies, and such information is generally not available unless the Pemafibrate cost distance between consecutive adjacencies is within the length of a read/fragment. In addition, the interval-adjacency graph does not contain allele-specific information about copy number variants, as considered in other work [35]. Finally, we assume that a cancer sample contains a single genome, when in fact most cancer samples contain DNA from a mixture of tumor cells, each with potentially different somatic mutations. It is possible that some of this intratumor heterogeneity could be resolved computationally. Alternatively,DNA sequencing of single cells, or smaller pools of cells, will minimize these effects. An additional area of investigation is to infer the temporal history of rearrangements. In the case of copy-neutral rearrangements, inferences can be made using parsimony models such as Hannenhalli-Pevzner theory [42]. This approach has previously been used in cancer genome analysis [13]. Models have also been introduced to infer orders of mutations in cases where there is interaction between duplications and rearrangements [43] and duplications and single-nucleotide mutations [35,44].Conclusions We formulated the Copy Number and Adjacency Genome Reconstruction Problem of reconstructing a rearranged cancer genome and developed an efficient algorithm, called Paired-end Reconstruction of Genome Organization (PREGO), for a particular instance of this problem. We designed an optimization problem on the interval-adjacency graph, which is related to the breakpoint graph used in genome rearrangement studies. We applied our algorithm to 5 ovarian cancer genomes sequenced as part of The Cancer Genome Atlas (TCGA) and reconstruct structural variants in these genomes. We analyzed the patterns of reciprocal vs. non-reciprocal rearrangements, and identified rearrangements consistent with known mechanisms of duplication such as tandem duplications and breakage/fusion/bridge cycles.Additional materialAdditional file 1: Figures of the interval-adjacency graph derived for all 5 ovarian genomes analyzed when cancer adjacencies Aare restricted to those with at leas.