Host innate immune response profiling reveals hidden viral infections across diverse animal species

Virus discovery using RNA-seq data from wildlife and livestock offers a powerful strategy for identifying unknown pathogens with pandemic potential. However, conventional approaches rely on homology-based searches that have limited sensitivity for highly divergent viruses, are computationally intensive at scale, and cannot distinguish true infections from contamination. Viral infection induces interferon-stimulated genes (ISGs), key components of the frontline antiviral defense, and their expression serves as a robust indicator of viral infection. Here, we developed a host-response–based virus discovery framework that rapidly quantifies ISG expression and predicts viral infection status. Applying this framework to ~210,000 RNA-seq data sets from diverse mammalian and avian species, we identified hidden viral infections across diverse hosts, including those caused by highly divergent viruses missed by a conventional approach. Our framework complements existing virus discovery strategies by adding host innate immune response context and enabling computationally efficient prescreening for scalable viral surveillance. Data sets Data set 1: Metadata for the Shaw et al. RNA-seq data set.Data set 2: Metadata for the He et al. and Zhao et al. RNA-seq data sets.Data set 3: Metadata for the 170k RNA-seq data set.Data set 4: Metadata for the 40k RNA-seq data set.Data set 5: ISG Profiler output for the Shaw et al. RNA-seq data set.Data set 6: ISG Profiler output for the He et al. and Zhao et al. RNA-seq data sets.Data set 7: ISG Profiler output for the 170k RNA-seq data set.Data set 8: ISG Profiler output for the 40k RNA-seq data set.Data set 9: Metadata for sequences in the ortholog database.Data set 10: geNomad results for the 170k RNA-seq data set.Data set 11: geNomad results for the 40k RNA-seq data set.Data set 12: Viral-derived contigs identified by BLASTx.Data set 13: Metadata for viral sequences used in the phylogenetic analyses.利用野生动物与家畜的RNA测序（RNA-seq）数据开展病毒发现研究，是鉴定具有大流行潜力的未知病原体的有效策略。然而，传统方法依赖基于同源性的搜索，这类方法对高度变异病毒的检测灵敏度有限，大规模运行时计算成本高昂，且无法区分真实感染与样本污染。病毒感染会诱导干扰素刺激基因（ISGs）的表达——这类基因是宿主一线抗病毒防御的核心组分，其表达水平可作为病毒感染的可靠检测指标。本研究开发了一种基于宿主应答的病毒发现框架，可快速定量干扰素刺激基因的表达水平并预测病毒感染状态。我们将该框架应用于来自多样哺乳动物与鸟类物种的约21万个RNA测序数据集，成功在多种宿主中鉴定出了隐藏的病毒感染，其中包括传统方法未能检出的高度变异病毒引发的感染。本框架通过引入宿主固有免疫应答背景信息，并实现了计算高效的预筛选以支持大规模病毒监测，从而对现有病毒发现策略形成有效补充。 数据集 数据集1：肖等人（Shaw et al.）RNA测序（RNA-seq）数据集的元数据。 数据集2：何等人与赵等人（He et al. and Zhao et al.）RNA测序（RNA-seq）数据集的元数据。 数据集3：17万份RNA测序数据集的元数据。 数据集4：4万份RNA测序数据集的元数据。 数据集5：肖等人RNA测序数据集的ISG Profiler分析输出结果。 数据集6：何等人与赵等人RNA测序数据集的ISG Profiler分析输出结果。 数据集7：17万份RNA测序数据集的ISG Profiler分析输出结果。 数据集8：4万份RNA测序数据集的ISG Profiler分析输出结果。 数据集9：直系同源数据库中序列的元数据。 数据集10：17万份RNA测序数据集的geNomad分析结果。 数据集11：4万份RNA测序数据集的geNomad分析结果。 数据集12：通过BLASTx鉴定得到的病毒来源重叠群。 数据集13：系统发育分析所用病毒序列的元数据。