Toward DNA Sequencing as a Primary Newborn Screen for Treatable Disorders not Amenable to Current Screening

我们建议评估全外显子组测序（WES）和全基因组测序 (WGS)作为对早发性可治疗的遗传性复发性疾病进行人群筛查的方法， 条件我们的分析将基于先前和额外的工作， 构成所有1，334例先天性代谢缺陷（IEM）的不同儿童队列 在加州通过串联质谱筛查诊断8.5年。 我们先前的分析发现，WES缺乏足够的敏感性和特异性来替代 目前新生儿筛查的串联质谱（MS/MS）。筛选优化 DNA变异解释管道在大多数情况下鉴定了两种已知或可能的致病性变异， 但不是所有受影响的病例。该管道的敏感性为88%，特异性为98.4%，数字太低 我们发现，1/3的已知或可能的致病变异是新的， 大概有三分之二是错的因此，适当的灵敏度和特异性要求 准确注释错义变体。 DNA变异注释集中在个体变异上，最适用于 致病性由单一变异体决定的显性疾病。对于隐性 在某些疾病中，表达由不同染色体上的两种变体决定。我们提出了一个 隐性疾病预测的新框架，其中双等位基因变体，或 双体型用于评估致病性。我们将改善常染色体疾病的检测 通过利用DNA测序筛查隐性疾病， 通过使用变体共进化、保守性、氨基酸邻近性 和其他特征来得出风险评分，2）考虑隐性致病性中的两种变体 预测，以及3）使用加州新生儿筛查数据集作为训练集， 每种疾病的阳性数量和大量的对照（那些对一种疾病呈阳性的病例）。 不同的疾病）来开发机器学习算法来预测疾病的可能性。到 地址敏感性，对于我们数据集中的103例外显子组阴性IEM病例，我们将充分询问 已知IEM基因和整个外显子组的外显子组数据，以鉴定新基因。 最后，对于那些尚未解决的问题，我们将执行WGS以确定其他因果关系 变体。在此过程中，我们还将比较WES和WGS的总体灵敏度， 筛选的特异性。 我们预计，这项研究将是评估DNA的一个重大进展。 测序作为新生儿筛查工具，用于那些早发性可治疗疾病， 目前没有筛查测试，导致死亡和残疾减少。

We propose to evaluate whole exome sequencing (WES) and whole genome sequencing (WGS) as an approach for population screening of early-onset treatable recessively inherited conditions. Our analysis will be based on prior and additional work on the large, ancestrally diverse cohort of children constituting all 1,334 cases of inborn errors of metabolism (IEM) diagnosed by tandem mass spectrometry screening in California over an 8.5 year period. Our prior analysis found that WES lacked adequate sensitivity and specificity to replace current newborn screening by tandem mass spectrometry (MS/MS). A screening-optimized DNA variant interpretation pipeline identified two known or likely pathogenic variants in most, but not all, affected cases. The pipeline was 88% sensitive and 98.4% specific, numbers too low to replace MS/MS. We found that 1/3 of known or likely pathogenic variants were novel, of which roughly 2/3 were missense. Adequate sensitivity and specificity therefore require accurate annotation of missense variants. DNA variant annotation has focused on individual variants and is most applicable for dominant disorders for which pathogenicity is determined by a single variant. For recessive diseases, expression is determined by two variants on different chromosomes. We propose a new framework for disease prediction in recessive conditions, in which the bi-allelic variants, or diplotype, is used to assess pathogenicity. We will improve disease detection for autosomal recessive diseases through screening utilizing DNA sequencing by 1) improving interpretation of bi-allelic missense variants by using variant co-evolution, conservation, amino acid proximity and other features to derive a risk score, 2) considering both variants in recessive pathogenicity prediction, and 3) using the California newborn screening data set as a training set with a small number of positives for each disorder and a large number of controls (those cases positive for a different disease) to develop a machine learning algorithm to predict likelihood of disease. To address sensitivity, for 103 exome negative IEM cases from our data set we will fully interrogate the exome data, both for known IEM genes and also the entire exome to identify novel genes. Finally, for those remaining unsolved, we will perform WGS to identify additional causal variants. In so doing, we will also compare WES and WGS in terms of overall sensitivity and specificity for screening. We anticipate that the proposed study will be a significant advance in assessing DNA sequencing as a newborn screening tool for those early-onset treatable diseases for which there is currently no screening test, leading to decreased death and disability.