Large-Scale Methods for Assessing the Consequences of Mutations in Proteins

摘要 每一种可能的与生命相容的错义变体都可能存在于活人的生殖系中。 这些变异中的一些会改变蛋白质的活性或丰度，因此可能会影响疾病风险。 然而，目前报道的所有错义突变中只有~2%具有临床解释。大多数 其余的变体以及几乎所有尚未观察到的错义变体都是罕见的，无法解释 使用传统方法，给基因组信息的临床应用带来了重大挑战。我们的目标是 通过衡量几乎所有可能的错义变体的功能后果来应对这一挑战 在临床相关蛋白质中使用深度突变扫描。在深度突变扫描中，一个蛋白质文库 对蛋白质的功能进行变异的选择，并使用高通量DNA测序 读出每个变异体的富集度或衰减值，揭示该变异体的功能。尽管最近 尽管如此，深度突变扫描面临着两个主要限制。第一个问题在于要求 为每种蛋白质的功能手工制作一种特定的分析方法。有超过4,000个疾病相关基因 人类基因组，这种一次一个的方法是不切实际的。因此，我们提出了大量的变异丰度。 平行测序(VAMP-SEQ)，一种既提供变异效应信息又可推广的功能分析 对许多蛋白质都有影响。这种分析是基于这样一个事实，尽管它们具有多样性，但大多数蛋白质共享一个关键 要求：它们必须足够丰富，才能发挥其分子功能。我们将生成VAMP-seq 一组10种临床重要蛋白质中几乎所有可能的错义变体的丰富数据，精炼 VAMP-SEQ作为一种工具，用于评估许多(如果不是大多数)疾病相关基因的错义变异。我们会 还将VAMP-SEQ与化学扰动相结合，以揭示蛋白质合成的基本特征， 折叠和降解，以及确定其低丰度可以改善的变体 从药理上讲。第二个主要限制是，深度突变扫描通常会量化 蛋白质活性或细胞生长的变异。这些简单的测量有时无法捕捉到 基因分型与人类表型关系的复杂性。因此，我们建议使用显微镜- 辅助视觉空间分类(MAVIS)，这将支持对更多 复杂的表型，如细胞的内部组织、形状或行为。我们将把Mavis应用于几个 疾病相关基因，为几乎所有可能的错义变异生成丰富的表型数据。我们的数据 来自VAMP-SEQ和MAVIS的集合将被用来生成描述以下内容的综合“查找表” 每个基因中几乎每一个错义变异的影响。我们还将分析这些不同的影响 已知的致病和良性变异的背景，使用基于学习的方法来全面 对错义变异致病性的预测。

ABSTRACT Every possible missense variant that is compatible with life is likely present in the germline of a living human. Some of these variants alter protein activity or abundance, and, consequently, may impact disease risk. However, only ~2% of all presently reported missense variants have clinical interpretations. Most of the remaining variants, as well as nearly all missense variants not yet observed, are rare and cannot be interpreted using traditional approaches, creating a major challenge for the clinical use of genomic information. Our goal is to address this challenge by measuring the functional consequences of nearly every possible missense variant in clinically relevant proteins using deep mutational scanning. In a deep mutational scan, a library of protein variants is subjected to selection for the function of the protein, and high-throughput DNA sequencing is used to read out the enrichment or depletion of each variant, revealing the variant's function. Despite recent progress, deep mutational scanning suffers from two major limitations. The first lies in the requirement to handcraft a specific assay for the function of each protein. With over 4,000 disease-associated genes in the human genome, this one-at-a-time approach is impractical. Thus, we propose Variant Abundance by Massively Parallel Sequencing (VAMP-seq), a functional assay that is both informative of variant effect and generalizable to many proteins. The assay is based on the fact that, despite their diversity, most proteins share a key requirement: they must be abundant enough to perform their molecular function. We will generate VAMP-seq abundance data for nearly all possible missense variants in a set of ten clinically important proteins, refining VAMP-seq as a tool for assessing missense variation in many, if not most, disease-relevant genes. We will also combine VAMP-seq with chemical perturbations to reveal fundamental features of protein synthesis, folding and degradation, as well as to identify variants whose low abundance could be ameliorated pharmacologically. The second major limitation is that deep mutational scans typically quantify the effect of variants on a protein's activity or on cell growth. These simple measurements sometimes fail to capture the complexity of the relationship between genotype and human phenotype. Thus, we propose Microscope- Assisted Visuospatial Sorting (MAViS), which will enable multiplex assessment of variant effects on more complex phenotypes like a cell's internal organization, shape or behavior. We will apply MAViS to several disease-related genes, generating rich phenotypic data for nearly all possible missense variants. The data we gather from both VAMP-seq and MAViS will be used to generate comprehensive “look-up tables” describing the effects of nearly every missense variant in each gene. We will also analyze these variant effects in the context of known pathogenic and benign variants, using a learning-based approach to make comprehensive predictions of missense variant pathogenicity.