Core 2: Bioinformatics and Biostatistics Core

核心 2：生物统计学和生物信息学 摘要/摘要 该卵巢孢子中的所有项目都使用各种测定法生成大量数据，并且 与测试新治疗方案疗效的临床试验相关。生物统计学和 生物信息学核心（核心 2）提供从这些信息中提取有用信息所需的定量专业知识 各种类型的数据，并设计和分析正在进行的临床试验的数据。这些都不是 任务是微不足道的。现已提供检测（例如，用于突变分析的下一代测序、甲基化 用于表观遗传学研究的阵列、表达阵列）非常强大，可以检测细微的变化 可能正在驱动表型变化，但这种敏感性意味着它们也很有能力检测 如果不使用正确的实验设计（例如随机化），则会出现测定伪影。此外，生成的原始数据 在得出有效的推论之前需要进行大量的预处理。 Core 2 成员接受过培训 为了解决这些不同项目所共有的任务，正在使用许多相同的检测方法 考虑到不同的实验目标，但数据分析问题是并行的。同样，虽然所有临床 试验有共同的目标（剂量发现、治疗评估），明确规定必须采取哪些措施 开展可能提供最多信息同时将患者面临的风险最少的试验需要专业知识 相关信息的获取和检查可能结果的能力（通常通过 模拟）。对于最常用的方法存在一些常见的设计，但最近的进展 过去几十年使我们能够制定可能更适合当前任务的新战略 现在存在计算能力来探索如何改进操作特性。 为项目提供最佳支持还需要灵活应对新的挑战和 可能出现的机会。在 SPORE 的这一迭代中，其中一些挑战和机遇 包括尝试最佳地利用 MD 安德森中心独有的 (a) 数据，其中耦合 腹腔镜检查和活检以及手术评估使我们能够了解治疗前和治疗后的情况 来自感兴趣人群的样本，以及 (b) 丰富的公共分析数据（例如 TCGA 卵巢检测） 这可以帮助从谷壳中筛选出真实的现象。 Core 2 具有这种灵活性，并且正在与 SPORE 合作 即使现在，调查人员仍在执行这些任务。 此外，正如 SPORE 中正在进行的研究表明，复杂分析和分析之间的分离 临床试验变得越来越模糊。向前迈进需要清楚地思考什么类型 推论可以可靠地使用，以及如何使用。 Core 2 成员因其对以下领域的贡献而受到广泛认可 这场辩论，这使他们能够更好地支持SPORE。

Core 2: Biostatistics and Bioinformatics SUMMARY/ABSTRACT All projects in this Ovarian SPORE generate large amounts of data using a wide variety of assays, and are associated with clinical trials testing the efficacy of new therapeutic options. The Biostatistics and Bioinformatics Core (Core 2) provides the quantitative expertise required to distill useful information from these various types of data, and to design and analyze the data from the clinical trials being run. Neither of these tasks is trivial. Assays now available (e.g., next-generation sequencing for mutation analysis, methylation arrays for epigenetic studies, expression arrays) are extremely powerful and can detect subtle changes that may be driving phenotypic changes, but this very sensitivity means they are also quite capable of detecting assay artifacts if proper experimental design (e.g., randomization) is not used. Further, the raw data generated requires substantial preprocessing before valid inferences can be drawn. Members of Core 2 have the training to address these tasks, which are common to the various projects—many of the same assays are being used with different experimental goals in mind, but the data analytic questions are parallel. Similarly, while all clinical trials have common goals (dose finding, treatment assessment), explicitly specifying what must be done to produce trials likely to provide the most information while putting the fewest patients at risk requires expertise in both the elicitation of relevant information and the ability to examine likely outcomes (often through simulation). Some common designs exist for the most frequent approaches pursued, but recent advances in the past few decades have allowed us to develop new strategies that may be better suited to the tasks at hand where the computing power now exists to explore how the operating characteristics are improved. Providing optimal support for the projects also requires the flexibility to address new challenges and opportunities that may arise. In this iteration of the SPORE, some of these challenges and opportunities include trying to optimally exploit both (a) data uniquely available within MD Anderson, where coupling of laparoscopic examination and biopsies with surgical evaluation lets us acquire both pre- and post-treatment samples from the population of interest, and (b) the wealth of public profiling data (e.g. TCGA ovarian assays) which can help winnow real phenomena from chaff. Core 2 has this flexibility, and is working with SPORE investigators on these tasks even now. Further, as the studies being pursued in the SPORE show, the separation between complex assays and clinical trials is becoming increasingly blurred. Moving forward will require clear thinking about what types of inferences can be reliably used, and how. Members of Core 2 are widely recognized for their contributions to this debate, which enables them to better support the SPORE.