Department of Epidemiology, Biostatistics and Occupational Health, McGill University, Montreal, Quebec, Canada.
Curr Opin Allergy Clin Immunol. 2013 Oct;13(5):470-7. doi: 10.1097/ACI.0b013e3283648f68.
During the past 2 years, next-generation sequencing studies have revolutionized the field of genetic association studies. We review the concomitant evolution of statistical methods.
As much of the genetic variability identified with sequencing is extremely rare, many new methods have been developed for rare variant association studies. Sequencing data available as a result of large public projects are also being integrated with genome-wide association study (GWAS) chip data to improve genotype imputation. A further trend in recent methodological development has been the use of the linear mixed effect model (LMM). LMMs are used for rare variant association to handle effect heterogeneity. They are also used more generally in GWAS to account for population structure.
Many rare variant association tests have been developed to analyze the genetic variation discovered with large-scale DNA sequencing; however, no single approach outperforms others under all disease models and power tends to be low. Sequencing data are also contributing to improved imputation of uncommon genetic variants, although imputation of rare variants remains a challenge. The appropriate correction for population structure in rare variant analyses remains unclear; specialized adjustment techniques may be necessary.
在过去的 2 年中，下一代测序研究彻底改变了遗传关联研究领域。我们回顾了同时发生的统计方法的演变。
由于测序所识别的大部分遗传变异都是极其罕见的，因此已经开发了许多用于罕见变异关联研究的新方法。由于大型公共项目而获得的测序数据也与全基因组关联研究 (GWAS) 芯片数据进行了整合，以改善基因型推断。最近方法发展的另一个趋势是使用线性混合效应模型 (LMM)。LMM 用于罕见变异关联，以处理效应异质性。它们也更广泛地用于 GWAS 以解释群体结构。
已经开发了许多罕见变异关联测试来分析大规模 DNA 测序发现的遗传变异；然而，在所有疾病模型下，没有一种单一的方法表现优于其他方法，而且功效往往较低。测序数据也有助于提高对不常见遗传变异的推断，尽管对稀有变异的推断仍然具有挑战性。在罕见变异分析中适当校正群体结构仍不清楚；可能需要专门的调整技术。
