Genome sequencing of large collections of individuals with rare diseases is accelerating the discovery of causal genes for monogenic disorders, but extracting robust associations between genotypes and phenotypes remains computationally and statistically challenging. To address this, Greene et al. developed a relational database framework called ‘Rareservoir’ that enables a more efficient representation of rare variants. Combining this framework with a Bayesian method to analyze matched genotype and phenotype data from 77,539 individuals sequenced by the 100,000 Genomes Project, they identified 260 significant associations between individual genes and specific disease groups, of which 241 were known and 19 were previously unidentified. Of the 19, the authors focused on the 3 associations with the highest probability scores, performed additional validation studies, including replication analyses in independent rare disease collections, and established convincing rare variant associations between ERG and primary lymphedema, PMEPA1 and Loeys–Dietz syndrome, and GPR156 and congenital hearing loss. The analytical approach used in this study showed high specificity for known associations and should aid the discovery of genetic associations for rare diseases as sample sizes for large-scale sequencing projects continue to grow.

对大量患有罕见疾病的个体进行基因组测序正在加速发现单基因疾病的致病基因，但提取基因型和表型之间的稳健关联在计算和统计学上仍然具有挑战性。为了解决这个问题，Greene 等人。开发了一个名为“Rareservoir”的关系数据库框架，可以更有效地表示稀有变体。将此框架与贝叶斯方法相结合，分析来自 100,000 人基因组计划测序的 77,539 名个体的匹配基因型和表型数据，他们确定了个体基因与特定疾病组之间的 260 种显着关联，其中 241 种是已知的，19 种以前未被识别。在 19 个中，作者关注概率得分最高的 3 个关联，ERG和原发性淋巴水肿、PMEPA1和 Loeys-Dietz 综合征，以及GPR156和先天性听力损失。本研究中使用的分析方法显示出对已知关联的高度特异性，并且随着大规模测序项目的样本量不断增加，应该有助于发现罕见疾病的遗传关联。