<b>G</b>ene-level <b>I</b>ntegrated <b>M</b>etric of negative <b>S</b>election (GIMS) Prioritizes Candidate Genes for Nephrotic Syndrome

Nephrotic syndrome (NS) gene discovery efforts are now occurring in small kindreds and cohorts of sporadic cases. Power to identify causal variants in these groups beyond a statistical significance threshold is challenging due to small sample size and/or lack of family information. There is a need to develop novel methods to identify NS-associated variants. One way to determine putative functional relevance of a gene is to measure its strength of negative selection, as variants in genes under strong negative selection are more likely to be deleterious. We created a gene-level, integrated metric of negative selection (GIMS) score for 20,079 genes by combining multiple comparative genomics and population genetics measures. To understand the utility of GIMS for NS gene discovery, we examined this score in a diverse set of NS-relevant gene sets. These included genes known to cause monogenic forms of NS in humans as well as genes expressed in the cells of the glomerulus and, particularly, the podocyte. We found strong negative selection in the following NS-relevant gene sets: (1) autosomal-dominant Mendelian focal segmental glomerulosclerosis (FSGS) genes (p= 0.03 compared to reference), (2) glomerular expressed genes (p = 4×10-23), and (3) predicted podocyte genes (p = 3×10-9). Eight genes causing autosomal dominant forms of FSGS had a stronger combined score of negative selection and podocyte enrichment as compared to all other genes (p=1 x 10-3). As a whole, recessive FSGS genes were not enriched for negative selection. Thus, we also created a transcript-level, integrated metric of negative selection (TIMS) to quantify negative selection on an isoform level. These revealed transcripts of known autosomal recessive disease-causing genes that were nonetheless under strong selection. We suggest that a filtering strategy that includes measuring negative selection on a gene or isoform level could aid in identifying NS-related genes. Our GIMS and TIMS scores are available at http://glom.sph.umich.edu/GIMS/.肾病综合征（NS）的基因发掘工作目前多开展于小型家系及散发病例队列中。由于样本量较小且/或缺乏家系信息，在此类研究中突破统计学显著性阈值以识别致病变异颇具挑战，因此亟需开发可用于识别NS相关致病变异的新型方法。判断某基因潜在功能相关性的一种可行方式，是检测其负选择（negative selection）强度——因为处于强负选择压力下的基因所携带的变异，更有可能具有致病性。本研究通过整合多项比较基因组学与群体遗传学检测指标，为20079个基因构建了基因水平负选择整合评分（GIMS）。为探究GIMS在NS基因发掘中的应用价值，本研究针对多组与NS相关的基因集开展了GIMS评分分析，涵盖的基因集包括已被证实可导致人类单基因型NS的基因，以及在肾小球细胞尤其是足细胞（podocyte）中表达的基因。本研究在以下与NS相关的基因集中观察到显著的负选择信号：(1) 常染色体显性遗传性局灶节段性肾小球硬化（FSGS）致病基因（与参照组相比，p=0.03）；(2) 肾小球表达基因（p=4×10^-23）；(3) 预测的足细胞（podocyte）特异性基因（p=3×10^-9）。与其余所有基因相比，8个常染色体显性FSGS致病基因同时具备更强的负选择评分与足细胞富集特征（p=1×10^-3）。整体而言，隐性FSGS致病基因未呈现负选择富集特征。据此，本研究进一步构建了转录本水平负选择整合评分（TIMS），以在异构体水平上量化负选择压力。该评分揭示了部分已知常染色体隐性致病基因的转录本：尽管其所属基因整体未呈现负选择富集，但自身转录本仍处于强负选择压力之下。本研究提示，纳入基因或异构体水平负选择检测的筛选策略，可辅助识别NS相关致病基因。本研究的GIMS与TIMS评分可通过以下网址获取：http://glom.sph.umich.edu/GIMS/