ERASE-Seq, an ultra-sensitive variant caller for liquid biopsy applications

摘要 基于cfDNA分析的液体活检测试已经引起了很多兴趣，并显示出能够 以确定接受癌症治疗的患者的体细胞突变状态。对于很大一部分 在晚期患者中，已经证明cfDNA和实体瘤变体之间的良好一致性。 尽管取得了这些令人兴奋的成果，但仍有很大的改进空间。即使使用前导cfDNA CLIA测试（Guardant Health）和检测晚期III期和IV期患者，约30%的样本没有 检测到足够高的流行率变体。这一领先的测试已证明检测限为 低至0.3%等位基因频率（Guardant Health），但其特异性仅被证明是高的 下降到2%。灵敏度和特异性的提高将使测试提供者和个体基因组 检测实验室在广泛的患者人群中常规使用液体活检检测， 过程 该项目的目标是使用ERASE-Seq开发液体活检工作流程，ERASE-Seq是一种超灵敏的体细胞生物学技术， 变体调用者。这是第一个基于样本间数据和背景数据进行变异调用的变异调用程序。 感知模型，消除了重复出现的工件。一个变量事件矩阵被组装，包含数量 每个变体在多个样品和背景运行中的出现次数。然后通过统计测试确定 每个变体相对于背景的存在的置信度得分，消除了导致 假阳性呼叫。 使用gDNA对各种不同的测序进行初始开发和验证。 技术.在测序数据中低至0.1%等位基因频率的变体的高灵敏度检测是可行的。 所示，在全靶向体细胞组（> 30 kb）中没有检测到假阳性。这一业绩 上级于目前可用的任何分子条形码面板和液体活检CLIA测试， 并且特异性数据可用。 如果成功，这笔赠款将开发一种用于cfDNA样本中变异检测的通用工具， 针对各种不同的工作流程和面板，灵敏度大幅提高，而不会增加复杂性 与分子条形码相关。所有用户将不得不使用这个工具是使用ERASE-Seq有针对性的 扩增试剂盒，并将fastq数据传送到我们的ERASE调用者软件。

Abstract Liquid biopsy tests based on cfDNA analysis have garnered a lot of interest, and show promise for being able to determine the somatic mutational status of patients undergoing cancer treatment. For a significant fraction of advanced stage patients, good agreement has been demonstrated between cfDNA and solid tumor variants. Despite these exciting results, significant room for improvement remains. Even when using the leading cfDNA CLIA test (Guardant Health) and testing advanced stage III and IV patients, about 30% of samples don’t have sufficiently high prevalence variants to be detected. This leading test has demonstrated a detection limit of as low as 0.3% allelic frequency (Guardant Health), but its specificity has only been demonstrated to be high down to 2%. Sensitivity and specificity improvements will enable both test providers and individual genomic test labs to routinely employ liquid biopsy tests across broad patient populations, and earlier in the disease process. The goal of this project is to develop a liquid biopsy workflow using ERASE-Seq, an ultra-sensitive somatic variant caller. This is the first variant caller to make variant calls based on intersample data and a background- aware model, which eliminates recurrent artifacts. A variant event matrix is assembled, containing the number of occurrences in multiple sample and background runs for each variant. A statistical test then determines the confidence score for presence of each variant over background, eliminating stochastic errors that contribute to false positive calls. The initial development and validation was performed using gDNA on a variety of different sequencing technologies. High sensitivity detection of variants in sequencing data down to 0.1% allelic frequency was shown, with no false positives detected across a full targeted somatic panel (>30kb). This performance is superior to any molecular barcoding panels currently available and liquid biopsy CLIA tests for which sensitivity and specificity data is available. If successful, this grant will develop a universal tool for variant detection in cfDNA samples, providing order of magnitude gains in sensitivity for a variety of different workflows and panels, without the added complications associated with molecular barcodes. All users will have to do to use this tool is use the ERASE-Seq targeted amplification kit and deliver fastq data to our ERASE caller software.