利用高层病因学知识从精神疾病全基因组关联研究中预测因果基因
全基因组关联研究已发现数百个与精神疾病特征相关的基因组位点,但其背后的因果基因常不明确,阻碍了临床转化。作者提出仅基于三个二值特征(突变约束、脑特异性表达、与已知神经发育疾病基因重叠)的精神疾病全位点优先级评分(PsyOPS),并显示其性能与依赖数万特征的领域无关方法相当。PsyOPS优先的基因更可能具有GWAS变异直接调控的汇聚证据(DNA环化实验与表达/剪接QTL图谱),并给出如精神分裂症GABBR1等距离先导变异数百kb的基因示例。
要点
- PsyOPS仅用三个二值特征即可达到与数万特征方法相当的因果基因优先级性能。
- PsyOPS优先基因更可能同时获得DNA环化与QTL的汇聚调控证据。
- 示例包括距离先导变异数百kb的GABBR1(精神分裂症)被三种证据共同优先。
核心结论
- 将高层病因学知识纳入GWAS位点因果基因预测具有价值,PsyOPS为实验验证精神疾病候选基因提供了资源。
研究设计与样本
- 研究基于精神疾病GWAS位点,构建仅含三个二值特征的精神疾病全位点优先级评分(PsyOPS),并与依赖数万特征的领域无关方法比较性能。
主要发现
- PsyOPS在因果基因优先级任务上达到与依赖数万特征的领域无关方法相当的性能。
- PsyOPS优先的基因比其他同一位点基因更可能具有GWAS变异直接调控的汇聚证据(DNA环化实验与表达或剪接QTL图谱)。
- 给出被PsyOPS、DNA环化和QTL三者共同优先的基因示例,如精神分裂症的GABBR1,距先导变异数百kb。
关键数值与推荐
- PsyOPS基于三个二值特征:因果精神疾病基因可能受突变约束、在脑内特异性表达、与已知神经发育疾病基因重叠。
临床怎么用
- 原文未据此给出临床推荐。
局限性
- 原文未报告局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
Genome-wide association studies have discovered hundreds of genomic loci associated with psychiatric traits, but the causal genes underlying these associations are often unclear, a research gap that has hindered clinical translation. Here, we present a Psychiatric Omnilocus Prioritization Score (PsyOPS) derived from just three binary features encapsulating high-level assumptions about psychiatric disease etiology - namely, that causal psychiatric disease genes are likely to be mutationally constrained, be specifically expressed in the brain, and overlap with known neurodevelopmental disease genes. To our knowledge, PsyOPS is the first method specifically tailored to prioritizing causal genes at psychiatric GWAS loci. We show that, despite its extreme simplicity, PsyOPS achieves state-of-the-art performance at this task, comparable to a prior domain-agnostic approach relying on tens of thousands of features. Genes prioritized by PsyOPS are substantially more likely than other genes at the same loci to have convergent evidence of direct regulation by the GWAS variant according to both DNA looping assays and expression or splicing quantitative trait locus (QTL) maps. We provide examples of genes hundreds of kilobases away from the lead variant, like GABBR1 for schizophrenia, that are prioritized by all three of PsyOPS, DNA looping and QTLs. Our results underscore the power of incorporating high-level knowledge of trait etiology into causal gene prediction at GWAS loci, and comprise a resource for researchers interested in experimentally characterizing psychiatric gene candidates.
引用
- Wainberg M, Merico D, Keller MC, et al. Predicting causal genes from psychiatric genome-wide association studies using high-level etiological knowledge. Mol Psychiatry. 2022. PMID: 35411039. doi:10.1038/s41380-022-01542-6
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。