Gene expression network analysis to identify candidate drivers of treatment resistance in glioblastoma multiforme

胶质母细胞瘤（GBM）是一种无法治愈的脑癌。尽管接受了手术和强化放化疗，但患者在确诊后平均存活了14个月，因为100%的肿瘤会复发。我们无法杀死这些肿瘤可能是因为它们由许多不同的癌细胞群组成，这些癌细胞群由独特的DNA和/或RNA谱定义，这使其中一些癌细胞具有抵抗治疗的能力。我们必须明确识别、描述并学习如何杀死这些耐药细胞亚群。为此，二级主管小组从接受标准治疗的患者中收集了数百例配对的原发性和复发性GBMs，并在大块组织和单细胞水平上对其进行分子分析。我们现在计划利用系统生物学方法充分利用这些数据集。该项目的目的是评估治疗期间GBM基因表达网络的扰动，以突出负责赋予或促进转录重编程的候选分子，使其更具治疗抗性。这将使用我们现有的数据，并辅以我们参与的一个全球联盟所产生的数据。我们有高覆盖率的RNA测序数据，来自对原发性和匹配复发性GBM肿瘤。这些数据是链定向的，包括mRNA和非编码转录物。我们已经对约80,000个已知和新基因进行了转录发现和定量基因表达，以及由此产生的转录本。这些数据可分别用于在原发性和复发性样本中创建共表达网络。主要主管先前开发了一种方法，可以使用多达6万个节点从相关性中创建基因表达网络。这样的网络将告诉我们每个肿瘤组的调控和功能途径。更重要的是，通过治疗的网络结构变化将突出与治疗耐药亚群的出现或增加有关的转录（重新）编程的候选主调控因子。将这些网络与每个节点（基因）的生物学数据及其相互作用叠加在一起，将使我们能够解释和确定在实验室和临床前环境中进行测试的治疗靶向机制。目标是：1。利用原发性和复发性GBM基因表达数据构建和优化共表达网络2。利用网络理论通过处理识别网络结构中的扰动3。将生物学信息分层到网络的节点上，以确定治疗驱动的转录（重）编程的候选主调节器

Glioblastoma (GBM) is an incurable brain cancer. Patients survive, on average, 14 months post-diagnosis, despite receiving surgery and intensive chemoradiation, because 100% of tumours grow back. Our inability to kill these tumours is likely because they consist of numerous distinct cancer cell populations defined by unique DNA and/or RNA profiles which confer some of them with the ability to resist treatment. It is these treatment resistant cells subsets that we must specifically identify, characterise and learn how to kill. To this end, the secondary supervisor's group has collected hundreds of paired primary and recurrent GBMs from patients that received standard treatment, and is molecularly profiling them at the bulk tissue and single cell level. We now plan to fully exploit these datasets using systems biology approaches. The aim of this project is to assess the perturbations across GBM gene expression networks during therapy to highlight candidate molecules responsible for conferring, or facilitating transcriptional reprogramming to, a more treatment resistant state. This will be done using our existing data, and supplemented with that being produced within a global consortium that we are part of.We have high coverage RNA sequencing data from pairs of primary and matched recurrent GBM tumours. These data are strand directional and include mRNA and non-coding transcripts. We have performed denovo transcript discovery and quantified gene expression for ~80,000 known and novel genes, as well as the resulting transcripts. These data can be used to create co-expression networks within primary and recurrent samples separately. The primary supervisor has previously developed an approach for creating gene expression networks from correlations using as many as 60,000 nodes. Such networks will inform us on the regulatory and functional pathways in each tumour group. More importantly, changes in network structure through therapy will highlight candidate master regulators of transcriptional (re)programming in relation to the emergence or increased prevalence of treatment resistant subpopulations. Overlaying these networks with biological data on each node (gene) and the interactions therein will enable us to interpret and identify therapeutically targetable mechanisms to be tested in laboratory and preclinical settings. The objectives are:1. Build and optimise co-expression networks using primary and recurrent GBM gene expression data2. Use network theory to identify perturbations in network structure through treatment3. Layer biological information onto the nodes of the network to identify candidate master regulators of therapy-driven transcriptional (re)programming