Integrating CNV analysis into a NextGen sequencing clinical analytics platform

拷贝数变异(CNV)是人类疾病的常见原因。截至今天，CNV最常见的 使用微阵列检测到的。然而，微阵列很昂贵，并且不能有效地检测CNV 低于1000个BP的范围。该项目的目的是在一个临床测试中同时检测点突变和CNV，而不需要 不得不利用微阵列。我们的目标是开发一种易于使用的解决方案，使临床医生和 研究人员在不需要生物信息学和脚本编写技能的情况下进行这种类型的高级分析。这个 由此产生的好处包括： 成本节约：通过消除对额外微阵列测试的需求，实验室将能够简化其 利用下一代测序(NGS)数据的分析工作流。 临床收益率：建议的解决方案将能够在低于1000 BP的范围内检测较小的CNV事件 它们仍然没有被微阵列检测到。这对临床医生来说至关重要，因为他们正在评估基因组 诊断目的。 易用性：建议的解决方案将嵌入到Golden Helix VarSeq产品中，即 旨在实现复杂的分析工作流，无需编写脚本或编程。简单化 CNV分析等高级工作流的使用至关重要，因为精准医疗正变得越来越 更加主流。该解决方案的简单性还将简化医疗保健培训 进入这一领域的专业人士。 出于我们的目的，我们可以将CNV定义为相对于人类参考的DNA的任何缺失或插入 大小为≥50bps的序列。短于50个碱基对的删除和插入是常见的，但通常可以 通过对NGS数据分析中使用的常规变种调用算法进行检测。CNV的大小可能会有所不同 最高可达几个兆基数。我们同样对检测数万个碱基或更大的CNV感兴趣 尺寸，因为我们热衷于尽可能扩大检测范围。 从NGS数据中检测CNV是目前人类遗传学研究的一个重要课题。不同的解决方案 以研究为导向的小组已经发展起来。我们将建立在已有的最佳解决方案的基础上 描述到目前为止，并使其商业上可用。大多数当前的解决方案使用的模型仅 能够合并单一证据指标。我们的CNV检测方法将改为使用 能够合并从参考集合导出的多个证据度量的概率模型 样本。这是一项关键改进，使我们的方法有别于 文学。此外，这项工作还将带来更高级的能力，如杂合性丢失的检测 (LOH)、拷贝中性LOH和单亲二体(UPD)事件。

Copy Number Variations (CNVs) are a common cause of disease in humans. As of today, CNVs are most often detected using microarrays. However, microarrays are expensive and are not able to effectively detect CNVs below 1K bp range. This project is designed to detect both point mutations and CNVs in one clinical test without having to utilize microarrays. It is our goal to develop an easy to use solution that allows clinicians and researchers to conduct this type of advanced analysis without requiring bioinformatics and scripting skills. The resulting benefits include:  Cost savings: By eliminating the need for additional microarray tests, labs will be able to streamline their analytics workflows utilizing NextGen Sequencing (NGS) data.  Clinical yield: The proposed solution will be able to detect smaller CNV events in the sub 1K bp range that remain undetected by microarrays. This is crucial for clinicians as they are evaluating a genome for diagnostic purposes.  Ease of Use: The proposed solution will be embedded in the Golden Helix VarSeq product that is designed to enable complex analytics workflows without the need to script or program. The simplification of advanced workflows such as the CNV analysis is crucial as precision medicine is becoming more and more mainstream. The simplicity of the solution will also streamline the training of healthcare professionals who are entering into this field. For our purpose, we may define CNVs as any deletion or insertion of DNA with respect to the human reference sequence of size ≥ 50 bps. Deletions and insertions shorter than 50 bp are common, but in general can be detected through routine variant calling algorithms used in the analysis of NGS data. CNVs may range in size up to several megabases. We are equally interested in detecting CNVs that are tens of kilobases or greater in size, as we are keen to extend the detection range as large as possible. CNV detection from NGS data is currently a key topic in human genetics. Different solutions from mostly research oriented groups have been developed. We will build upon the best of the solutions that have been described to date and make them commercially available. Most current solutions use models that are only capable of incorporating a single evidence metric. Our approach to CNV detection will instead make use of a probabilistic model capable of incorporating multiple evidence metrics derived from a collection of reference samples. This is a key improvement that differentiates our approach from existing methods described in the literature. Also, this work will lead to more advanced capabilities such as the detection of loss of heterozygosity (LOH), copy-neutral LOH and uniparental disomy (UPD) events.