Department of Mathematics, London School of Economics and Political Science, London, UK.
Department of Computer Science, Helsinki Institute for Information Technology HIIT, University of Helsinki, Finland.
Bioinformatics. 2019 Mar 1;35(5):769-777. doi: 10.1093/bioinformatics/bty683.
Discovering the evolution of a tumor may help identify driver mutations and provide a more comprehensive view on the history of the tumor. Recent studies have tackled this problem using multiple samples sequenced from a tumor, and due to clinical implications, this has attracted great interest. However, such samples usually mix several distinct tumor subclones, which confounds the discovery of the tumor phylogeny.
We study a natural problem formulation requiring to decompose the tumor samples into several subclones with the objective of forming a minimum perfect phylogeny. We propose an Integer Linear Programming formulation for it, and implement it into a method called MIPUP. We tested the ability of MIPUP and of four popular tools LICHeE, AncesTree, CITUP, Treeomics to reconstruct the tumor phylogeny. On simulated data, MIPUP shows up to a 34% improvement under the ancestor-descendant relations metric. On four real datasets, MIPUP's reconstructions proved to be generally more faithful than those of LICHeE.
MIPUP is available at https://github.com/zhero9/MIPUP as open source.
Supplementary data are available at Bioinformatics online.
发现肿瘤的进化可能有助于鉴定驱动突变，并更全面地了解肿瘤的历史。最近的研究使用从肿瘤中测序的多个样本来解决这个问题，由于临床意义，这引起了极大的兴趣。然而，这些样本通常混合了几个不同的肿瘤亚克隆，这使得肿瘤系统发生的发现变得复杂。
我们研究了一个需要将肿瘤样本分解为几个亚克隆的自然问题，其目的是形成一个最小的完美系统发生树。我们为此提出了一种整数线性规划（Integer Linear Programming）的公式，并将其实现为一种名为 MIPUP 的方法。我们测试了 MIPUP 和四个流行工具 LICHeE、AncesTree、CITUP 和 Treeomics 重建肿瘤系统发生的能力。在模拟数据上，MIPUP 在祖先-后代关系度量上的改进高达 34%。在四个真实数据集上，MIPUP 的重建结果通常比 LICHeE 的更准确。
MIPUP 可在 https://github.com/zhero9/MIPUP 上作为开源获取。
补充数据可在 Bioinformatics 在线获取。
