Lorincz-Comi Noah, Yang Yihe, Ajayakumar Jayakrishnan, Mews Makaela, Bermudez Valentina, Bush William, Zhu Xiaofeng
Case Western Reserve University Department of Population and Quantitative Health Sciences, Cleveland, OH 44106, United States.
Case Western Reserve University Department of Neurosciences, Cleveland, OH 44106, United States.
Bioinform Adv. 2025 Apr 18;5(1):vbaf068. doi: 10.1093/bioadv/vbaf068. eCollection 2025.
Nearly two decades of genome-wide association studies (GWAS) have identify thousands of disease-associated genetic variants, but very few genes with evidence of causality. Recent methodological advances demonstrate that Mendelian randomization (MR) using expression quantitative loci (eQTLs) as instrumental variables can detect potential causal genes. However, existing MR approaches are not well suited to handle the complexity of eQTL GWAS data structure and so they are subject to bias, inflation, and incorrect inference.
We present a whole-genome regulatory network analysis tool (HORNET), which is a comprehensive set of statistical and computational tools to perform genome-wide searches for causal genes using summary level GWAS data, i.e. robust to biases from multiple sources. Applying HORNET to schizophrenia, eQTL effects in the cerebellum were spread throughout the genome, and in the cortex were more localized to select loci.
Freely available at https://github.com/noahlorinczcomi/HORNET or Mac, Windows, and Linux users.
近二十年的全基因组关联研究（GWAS）已经鉴定出数千种与疾病相关的遗传变异，但只有极少数基因有因果关系的证据。最近的方法学进展表明，使用表达定量基因座（eQTL）作为工具变量的孟德尔随机化（MR）可以检测潜在的因果基因。然而，现有的MR方法不太适合处理eQTL GWAS数据结构的复杂性，因此容易受到偏差、膨胀和错误推断的影响。
我们提出了一种全基因组调控网络分析工具（HORNET），它是一套全面的统计和计算工具，用于使用汇总水平的GWAS数据进行全基因组因果基因搜索，即对多种来源的偏差具有鲁棒性。将HORNET应用于精神分裂症，小脑的eQTL效应分布在整个基因组中，而在皮质中则更局限于选定的基因座。
可在https://github.com/noahlorinczcomi/HORNET上免费获取，适用于Mac、Windows和Linux用户。
