Human pathways in animal models: possibilities and limitations

Abstract

Animal models are crucial for advancing our knowledge about the molecular pathways involved in human diseases. However, it remains unclear to what extent tissue expression of pathways in healthy individuals is conserved between species. In addition, organism-specific information on pathways in animal models is often lacking. Within these limitations, we explore the possibilities that arise from publicly available data for the animal models mouse, rat, and pig. We approximate the animal pathways activity by integrating the human counterparts of curated pathways with tissue expression data from the models. Specifically, we compare whether the animal orthologs of the human genes are expressed in the same tissue. This is complicated by the lower coverage and worse quality of data in rat and pig as compared to mouse. Despite that, from 203 human KEGG pathways and the seven tissues with best experimental coverage, we identify 95 distinct pathways, for which the tissue expression in one animal model agrees better with human than the others. Our systematic pathway-tissue comparison between human and three animal modes points to specific similarities with human and to distinct differences among the animal models, thereby suggesting the most suitable organism for modeling a human pathway or tissue.

Graphical Abstract

Pathway-tissue comparison between human and three animal models.

Figure 1.

Transferability of 205 KEGG pathways from human to mouse, rat, and pig. Each bar represents the number of pathways, for which a given fraction of genes can be transferred. Transferability to each organism is shown in different colors: mouse in green, rat in red, pig in orange and all organisms in grey.

Figure 2.

Principal component analysis of the pathway–tissue agreement between human and animal models. PCA was performed on the Jaccard indices (JIs) for all pathway–tissue pairs (grey dots) with at least five expressed pathway genes, where the JIs represent the comparisons human–mouse, human–rat, and human–pig. The PCA loadings are shown as solid lines and colored by the model organism responsible for their direction. While the PC1 & PC2 plot (panel A) shows a clear separation between pathway–tissue pairs with high JI and thus, good agreement between human and the respective animal model, the PC2 & PC3 plot (panel B) clearly separates the data based on the differences between the animal models. The pathway–tissue pairs located closest to each loading and furthest away from the center of the PC2 & PC3 plot are colored in the same color as the organism loading to indicate that for these pairs, this organism agrees more with human than the others (see Methods section for more details).