DNA sequencing is a widely used tool in genetic association study. Sequencing cost remains a major concern in sequencing-based study, although the application of next generation sequencing has dramatically decreased the sequencing cost and increased the efficiency. The choice of sequencing depth and the sequencing sample size will largely determine the final study investment and performance. Many studies have been conducted to find a cost-effective design of sequencing depth that can achieve certain sequencing accuracy using minimal sequencing cost. The strategies previously studied can be classified into two groups: (1) single-stage to sequence all the samples using either high (>~30×) or low (<~10×) sequencing depth; and (2) two-stage to sequence an affordable number of individuals at a high-coverage followed by a large sample of low-coverage sequencing. However, limited studies examined the performance of the medium-coverage (10–30×) sequencing depth for a genetic association study, where the optimum sequencing depth may exist. In this study, using a published simulation framework, we comprehensively compared the medium-coverage sequencing (MCS) to the single- and two-stage high/low-coverage sequencing in terms of the power and type I error of the variant discovery and association testing. We found, given certain sequencing effort, MCS yielded a comparable discovery power and better type I error control compared with the best (highest power) scenarios using other high- and low-coverage single-stage or two-stage designs. However, MCS was not as competent as other designs with respect to the association power, especially for the rare variants and when the sequencing investment was limited.

DNA测序是遗传关联研究中广泛使用的工具。测序成本仍然是基于测序的研究中的一个主要问题，尽管二代测序的应用大大降低了测序成本并提高了效率。测序深度和测序样本量的选择将在很大程度上决定最终的研究投资和绩效。已经进行了许多研究以寻找一种经济有效的测序深度设计，该设计可以使用最小的测序成本实现一定的测序精度。先前研究的策略可以分为两组：（1）使用高（>~30×）或低（<~10×）测序深度对所有样本进行单阶段测序；(2) 两阶段以高覆盖率对负担得起的数量的个体进行测序，然后是大样本的低覆盖率测序。然而，有限的研究检查了中等覆盖（10-30×）测序深度在遗传关联研究中的性能，其中可能存在最佳测序深度。在本研究中，我们使用已发布的模拟框架，在变异发现和关联的功效和 I 型错误方面，全面比较了中等覆盖率测序 (MCS) 与单阶段和两阶段高/低覆盖率测序测试。我们发现，与使用其他高覆盖率和低覆盖率的单级或两级设计的最佳（最高功率）方案相比，在进行一定的测序工作时，MCS 产生了可比的发现能力和更好的 I 型错误控制。然而，