[论文解读] Finding Mutated Subnetworks Associated with Survival in Cancer

Tommy Hansen, Fabio Vandin|arXiv (Cornell University)|Apr 8, 2016

Cancer Genomics and Diagnostics参考文献 20被引用 6

一句话总结

该论文提出 NoMAS，一种新颖算法，利用颜色编码技术在基因互作网络中识别出与癌症患者生存显著相关的微小突变子网络。尽管基于 log-rank 检验的评分具有非加性特性，NoMAS 在合理突变模型下仍以高概率识别出最优子网络，并在模拟数据和 TCGA 癌症数据上均优于单基因检验和贪心方法，即使单个基因不显著，也能检测到与生存相关的子网络。

ABSTRACT

Next-generation sequencing technologies allow the measurement of somatic mutations in a large number of patients from the same cancer type. One of the main goals in analyzing these mutations is the identification of mutations associated with clinical parameters, such as survival time. This goal is hindered by the genetic heterogeneity of mutations in cancer, due to the fact that genes and mutations act in the context of pathways. To identify mutations associated with survival time it is therefore crucial to study mutations in the context of interaction networks. In this work we study the problem of identifying subnetworks of a large gene-gene interaction network that have mutations associated with survival. We formally define the associated computational problem by using a score for subnetworks based on the test statistic of the log-rank test, a widely used statistical test for comparing the survival of two populations. We show that the computational problem is NP-hard and we propose a novel algorithm, called Network of Mutations Associated with Survival (NoMAS), to solve it. NoMAS is based on the color-coding technique, that has been previously used in other applications to find the highest scoring subnetwork with high probability when the subnetwork score is additive. In our case the score is not additive; nonetheless, we prove that under a reasonable model for mutations in cancer NoMAS does identify the optimal solution with high probability. We test NoMAS on simulated and cancer data, comparing it to approaches based on single gene tests and to various greedy approaches. We show that our method does indeed find the optimal solution and performs better than the other approaches. Moreover, on two cancer datasets our method identifies subnetworks with significant association to survival when none of the genes has significant association with survival when considered in isolation.

研究动机与目标

• 为解决在癌症中识别与患者生存相关的突变子网络的挑战，克服因遗传异质性导致的单基因分析局限性。
• 开发一种整合体细胞突变数据、全基因组蛋白质-蛋白质互作网络和右删失生存数据的方法，以检测具有生物学意义的子网络。
• 设计一种高效算法，即使评分函数具有非加性特性，也能识别出与生存具有强统计关联的高分子网络。
• 证明该方法识别出的子网络可揭示在单基因水平上无法检测到的临床相关关联。

提出的方法

• NoMAS 采用颜色编码技术，在大型基因-基因互作网络中高效搜索大小最多为 8 的高分子网络。
• 该算法使用基于 log-rank 检验统计量的非加性评分来评估子网络与生存的关联性。
• 证明在合理的癌症突变模型下，尽管评分具有非加性，NoMAS 仍能以高概率识别出最优子网络。
• 该方法使用置换检验估计 p 值，并评估识别出的子网络的统计显著性。
• NoMAS 在模拟数据和来自 The Cancer Genome Atlas (TCGA) 的真实癌症数据集上进行了评估。
• 提出一种局部搜索策略，通过将高分小子网络作为种子顶点，将方法扩展至更大子网络。

实验结果

研究问题

• RQ1 计算方法能否在单个基因无显著关联的情况下，识别出与患者生存显著相关的突变基因子网络？
• RQ2 颜色编码技术能否在生存分析背景下，有效适配非加性评分函数？
• RQ3 NoMAS 在检测与生存相关的子网络方面，与单基因检验和贪心子网络识别方法相比表现如何？
• RQ4 当单基因分析失败时，NoMAS 能否在真实癌症数据集中检测到已知和新型预后子网络？
• RQ5 在现实突变模型下，NoMAS 识别最优子网络的可扩展性和准确性如何？

主要发现

• NoMAS 在两个 TCGA 癌症数据集中成功识别出与生存显著相关的子网络，即使子网络中无任何单个基因在孤立分析中具有显著性。
• 在模拟数据中，NoMAS 的运行时间显著低于穷举枚举，同时接近最优性能，保持了高准确性。
• 该算法识别出已知预后基因和新型候选基因，且这些子网络具有统计显著性，显示出生物学相关性。
• NoMAS 在检测与生存关联更强的子网络方面，优于单基因检验和多种贪心方法。
• 尽管 log-rank 评分具有非加性，NoMAS 在合理突变模型下仍能以高概率识别出最优子网络。
• 通过利用小子网络的高分子网络作为种子顶点，局部搜索扩展策略实现了对更大子网络的高效探索，表明未来可扩展的策略。