Multi-Layer Transcriptomic Comparison Across Human, Pig, and Mouse Informs Tissue- and Trait-Specific Animal Model Selection.

Animal models are fundamental to biomedical research, yet the relative suitability of pigs and mice remains unclear across tissues, regulatory layers, and phenotype domains. Here, we assembled a large-scale, tissue-matched transcriptomic resource comprising more than 16 000 RNA-seq samples from 15 corresponding human, pig, and mouse tissues, and established a unified framework integrating global co-expression conservation, co-expression network architecture, tissue-specific programs, 3D genome architecture, and human complex-trait genetics. Using 14 910 one-to-one orthologous genes, we found that human-pig pairs showed higher global co-expression similarity than human-mouse pairs in most tissues, with exceptions including skin, pancreas, and kidney. Integration with 176 human complex traits further revealed distinct phenotypic associations of species-preferred patterns: human-pig-preferred genes showed preferential associations with cardiometabolic, biochemical, and blood-related traits, whereas both-conserved genes showed stronger associations with neuropsychiatric, cognitive, and behavioral phenotypes. Together, our results demonstrate that cross-species similarity is hierarchical and context-dependent, varying across tissues, regulatory layers, and trait domains. Rather than supporting a universally optimal model organism, our study provides a comparative framework for prioritizing animal models based on regulatory conservation, together with an interactive portal for gene-centered exploration across humans, pigs, and mice.