Integrated modeLs for Early Risk-prediction in Africa (ILERA) study

项目摘要 心脏代谢疾病（CMD）每年在非洲夺走数百万人的生命，其中相当大的一部分 过早死亡。尽管有简单和负担得起的方法，如生活方式调整， 以及使用可以延长寿命和改善生活质量的药物（例如降脂他汀类药物）， 随着时间的推移，在非洲正成为一个越来越严重的健康负担。优先考虑人口的医疗保健的能力 在资源有限的环境中可能特别相关。的一个主要 根据风险对人群进行准确分层的挑战是目前多基因风险评分的预测性较低 在非洲人口中的模型（PRS）。 非洲早期风险预测综合模型（ILERA）研究（Ilera在Yourba的意思是健康）旨在 研究改善13种心脏代谢疾病指标水平预测的潜力（以及 通过将不同类型的数据（基因组学、转录组学、生活方式相关数据）整合到风险中， 预测模型从目前表现最好的PRS开始，我们计划逐步添加数据层 例如预测的转录组、环境和生活方式信息，以评估这些额外的数据， 无论是单独使用还是与其他方法结合使用，都可以改善预测。以允许复杂的和非线性的 这些因素之间的相互作用，数据驱动的方法将被用来整合这些变量， 基因组数据。将在独立队列中对这些模型的预测性进行深入评价 来自南部、东部和西部非洲的数据以及来自同一队列的纵向数据。早期的可能性 针对公共卫生干预的预警系统将使用最佳预测组合进行调查， 模型和特征。 该项目将由威特沃特斯兰德大学（Wits）领导，与Wits Donald Gordon合作 医疗中心，非洲生物医学科学和技术研究所（ABiST）津巴布韦和美国 基于行业合作伙伴，变异生物。预测的转录组将基于750名南非参与者 全基因组序列和血液转录组RNA-Seq。约5000的主要目标数据集 参与者是通过H3 Africa AWI-Gen研究产生的，这些模型将在两个南方国家进行测试。 非洲数据集（约1200名来自南非和津巴布韦的参与者）以及约6000名来自 加纳、布基纳法索和肯尼亚。将使用基线数据收集后5年采集的纵向数据 了解年龄对预测模型的影响。这项研究将建立在多年来现有的成功的 合作，并将利用Wits在基因组学研究方面的经验， 研究并利用与DSI-Africa联盟中其他项目的伙伴关系，以提高数据科学能力。

Project Summary Cardiometabolic diseases (CMDs) claim millions of lives in Africa every year and a sizable portion of these deaths are premature. Despite the availability of simple and affordable approaches such as lifestyle adjustment and the use of drugs (e.g. lipid lowering statins) that could increase lifespan and improve the quality of life, this is becoming a more serious health burden in Africa with time. The ability to prioritize healthcare to the populations that are at highest risk could be especially relevant in resource constrained environments. One of the major challenges to accurately stratifying a population by risk is the low predictivity of current polygenic risk scoring models (PRSs) in African populations. The Integrated modeLs for Early Risk-prediction in Africa (ILERA) study (Ilera in Yourba means health) aims to investigate the potential for improving the prediction of 13 cardiometabolic disease indicator levels (and thereby of CMDs) by integrating diverse types of data (genomic, transcriptomic, lifestyle-related data) into risk prediction models. Starting with currently best performing PRSs, we plan to progressively add layers of data such as predicted transcriptomes, environment and lifestyle information to assess whether this additional data, either independently or in combination with others, could improve prediction. To allow for complex and non-linear interactions between these factors, data-driven approaches will be employed to integrate these variables with the genomic data. In-depth evaluation of the predictivity of these models will be performed in independent cohorts from South, East and West Africa and also in longitudinal data from the same cohort. The potential for an early warning system aimed at public health intervention will be investigated using a combination of the best predictive models and traits. The project will be led from the University of the Witwatersrand (Wits), collaborating with the Wits Donald Gordon Medical Center, the African Institute of Biomedical Science and Technology (ABiST) Zimbabwe and an US based industry partner, Variant Bio. The predicted transcriptome will be based on 750 South African participants with whole genome sequence and blood transcriptome RNA-Seq. The primary target dataset of ~5000 participants was generated through the H3Africa AWI-Gen study and the models will be tested in two Southern African datasets (~1200 participants from South Africa and Zimbabwe) as well as ~6000 participants from Ghana, Burkina Faso and Kenya. Longitudinal data, captured 5 years after baseline data collection, will be used to understand the impact of age on the predictive models. The study will build on years of existing successful collaboration and will tap into the Wits experience in genomics research, Variant Bio’s expertise in multi-omics research and leverage partnership with other projects in the DSI-Africa consortium for data science capacity.