Li Xiaoyang, Fernandes Brisa S, Liu Andi, Chen Jingchun, Chen Xiangning, Zhao Zhongming, Dai Yulin
Center for Precision Health, McWilliams School of Biomedical Informatics, The University of Texas Health Science Center at Houston, Houston, TX 77030, USA.
Department of Biostatistics and Data Science, School of Public Health, The University of Texas Health Science Center at Houston, Houston, TX 77030, USA.
medRxiv. 2024 Jul 5:2023.06.19.23291621. doi: 10.1101/2023.06.19.23291621.
Polygenic risk scores (PRS) are tools used to evaluate an individual's susceptibility to polygenic diseases based on their genetic profile. A considerable proportion of people carry a high genetic risk but evade the disease. On the other hand, some individuals with a low risk of eventually developing the disease. We hypothesized that unknown counterfactors might be involved in reversing the PRS prediction, which might provide new insights into the pathogenesis, prevention, and early intervention of diseases.
We built a novel computational framework to identify genetically-regulated pathways (GRPas) using PRS-based stratification for each cohort. We curated two AD cohorts with genotyping data; the discovery (disc) and the replication (rep) datasets include 2722 and 2854 individuals, respectively. First, we calculated the optimized PRS model based on the three recent AD GWAS summary statistics for each cohort. Then, we stratified the individuals by their PRS and clinical diagnosis into six biologically meaningful PRS strata, such as AD cases with low/high risk and cognitively normal (CN) with low/high risk. Lastly, we imputed individual genetically-regulated expression (GReX) and identified differential GReX and GRPas between risk strata using gene-set enrichment and variational analyses in two models, with and without effects. An orthogonality test was further conducted to verify those GRPas are independent of PRS risk. To verify the generalizability of other polygenic diseases, we further applied a default model of GRPa-PRS for schizophrenia (SCZ).
For each stratum, we conducted the same procedures in both the disc and rep datasets for comparison. In AD, we identified several well-known AD-related pathways, including amyloid-beta clearance, tau protein binding, and astrocyte response to oxidative stress. Additionally, we discovered resilience-related GRPs that are orthogonal to AD PRS, such as the calcium signaling pathway and divalent inorganic cation homeostasis. In SCZ, pathways related to mitochondrial function and muscle development were highlighted. Finally, our GRPa-PRS method identified more consistent differential pathways compared to another variant-based pathway PRS method.
We developed a framework, GRPa-PRS, to systematically explore the differential GReX and GRPas among individuals stratified by their estimated PRS. The GReX-level comparison among those strata unveiled new insights into the pathways associated with disease risk and resilience. Our framework is extendable to other polygenic complex diseases.
多基因风险评分（PRS）是用于根据个体基因概况评估其对多基因疾病易感性的工具。相当一部分人携带高遗传风险但未患该疾病。另一方面，一些个体最终患该疾病的风险较低。我们推测可能存在未知的抵消因素参与逆转PRS预测，这可能为疾病的发病机制、预防和早期干预提供新的见解。
我们构建了一个新颖的计算框架，使用基于PRS的分层为每个队列识别基因调控通路（GRPas）。我们整理了两个有基因分型数据的AD队列；发现（disc）和复制（rep）数据集分别包括2722名和2854名个体。首先，我们基于每个队列的三个近期AD全基因组关联研究（GWAS）汇总统计数据计算优化的PRS模型。然后，我们根据个体的PRS和临床诊断将其分层为六个具有生物学意义的PRS层，如低/高风险的AD病例以及低/高风险的认知正常（CN）个体。最后，我们估算个体基因调控表达（GReX），并在两个模型（有和无效应）中使用基因集富集和变分分析识别风险层之间的差异GReX和GRPas。进一步进行正交性检验以验证这些GRPas独立于PRS风险。为了验证其他多基因疾病的可推广性，我们进一步将GRPa - PRS的默认模型应用于精神分裂症（SCZ）。
对于每个层，我们在disc和rep数据集中都进行了相同的程序以进行比较。在AD中，我们识别出了几个与AD相关的知名通路，包括淀粉样β清除、tau蛋白结合以及星形胶质细胞对氧化应激的反应。此外，我们发现了与AD PRS正交的与恢复力相关的GRPs，如钙信号通路和二价无机阳离子稳态。在SCZ中，突出了与线粒体功能和肌肉发育相关的通路。最后，与另一种基于变异的通路PRS方法相比，我们的GRPa - PRS方法识别出了更一致的差异通路。
我们开发了一个框架GRPa - PRS，以系统地探索根据估计的PRS分层的个体之间的差异GReX和GRPas。这些层之间的GReX水平比较揭示了与疾病风险和恢复力相关通路的新见解。我们的框架可扩展到其他多基因复杂疾病。
