(PQB5) Reconstruction of Evolutionary Networks using Cross-Sectional Genomic Data

描述(由申请人提供)：癌症是一个动态的过程，通过基因组变化的积累进行。大型测序项目已经照亮了大量肿瘤变化的复杂静态图景。然而，这些研究未能解决癌症的动态性质。了解肿瘤是如何被选择性压力塑造的，对治疗和预后具有重要意义。为了解决这个问题，我们提出了Pitch(简约推断肿瘤克隆异质性)，这是一个旨在利用横断面研究的高通量基因组数据揭示肿瘤进化历史的计算模型。螺距识别较老克隆的痕迹，并重建可能的病损历史。通过结合来自不同患者的数据，Pitch能够捕捉到肿瘤司机变化之间在统计上强大的历史关系，并将这些关系表示为一个进化网络。我们将在跨越12年的近1500名慢性淋巴细胞白血病患者和20例多形性胶质母细胞瘤样本的纵向队列中校准该方法。我们的目标是使用癌症基因组图谱中收集的大量多形性胶质母细胞瘤和低级别胶质瘤来扩展并实验验证该方法。我们将能够提供一种强大的计算方法，可以很容易地扩展到任何其他类型的肿瘤，在那里可以获得大量的横断面数据。发现与不同阶段、阶段或治疗策略相关的改变可以为基于基因组数据的个性化方法提供宝贵的生物标志物。

DESCRIPTION (provided by applicant): Cancer is a dynamic process that proceeds through the accumulation of genomic alterations. Large sequencing projects have illuminated the complex static landscapes of alterations across a large number of tumors. However, these studies have failed to address the dynamic nature of cancers. Understanding how tumors are shaped by selective pressures bears implications in therapies and prognoses. To address this issue we propose PITCH (Parsimony Inference of Tumor Clone Heterogeneity), a computational model that aims to uncover the evolutionary history of tumors using high throughput genomic data from cross-sectional studies. PITCH identifies traces of older clones and reconstructs possible histories of lesions. By combining the data from different patients, PITCH is able to capture statistically robust historical relationships between driver alterations in tumors and to represent these relationships as an evolutionary network. We will calibrate the approach in a longitudinal cohort of nearly 1,500 Chronic Lymphocytic Leukemia patients spanning a period of 12 years, along with 20 Glioblastoma Multiforme samples. We aim to extend and thus experimentally validate the approach using the large collection of Glioblastoma Multiforme and Low Grade Glioma in The Cancer Genome Atlas. We will be able to provide a robust computational approach that can be easily extended to any other tumor type where large cross-sectional data is available. Discovery of alterations associated to different phases, stages or therapeutic strategies could provide invaluable biomarkers for personalized approaches based on genomic data.