The Interaction ResourceProtein-protein interactions are essential for many processes in cellular life, such as cell signaling, gene expression and control and antibody–antigen binding, and they are often blocked or altered in disease. Knowledge about interaction networks can therefore help in understanding disease mechanisms and in developing diagnostic and therapeutic strategies. The Interaction resource presents interaction networks integrated with data related to protein expression, location and classification based on experimental protein-protein interaction data from IntAct, BioGRID, BioPlex and OpenCell for 15294 genes as well as high-confidence predictions based on RoseTTAFold2-PPI for 9540 genes, of which 1041 genes only has predicted data. For each gene seven interaction networks are displayed, one for each of the five individual datasets, one with aggregated data showing interactions from all datasets and one consensus network showing the interactions present in more than one experimental dataset. For 7298 genes and 20163 of their consensus interactions, interactive 3D protein structures based on predictions generated using the AlphaFold 3 source code are also displayed. More information about the specific content and the generation of the data in this section can be found in the Methods summary. Learn about:
The protein-protein interaction network plots based on data from IntAct, BioPlex, OpenCell, BioGRID and RosettaFold2-PPI high precision predictions enables exploration of the interaction partners in each dataset separately, in a network based on aggregation of all datasets and, if applicable, in a consensus network showing only interactions present in more than one of the experimental datasets. More information about the genes in the network is displayed by mouse-over and the network can be extended by clicking on any gene of interest.
Figure 1. First-level consensus interactions for the nuclear protein CCNE1 colored according to subcellular location (left) and the aggregated interaction data from all datasets for the CD marker CD28 colored according to cell type specificity (right). The edge colors represent the number of datasets for the interacting pair and datasets involved are shown by mouse-over. Dashed line edges represent interaction pairs based only on predicted data from RosettaFold2-PPI and a square on an edge highlights a known heterodimer. By clicking on the purple points on the edges the corresponding 3D structure for the interaction together with scores and and a PAE plot is displayed as shown in Figure 2. To allow for deeper understanding of the interaction partners and their context the genes in a network can be colored according to experimental subcellular location based on data in the Subcellular resource, predicted location based on signal peptide and transmembrane region predictors, tissue specificity based on RNA tissue expression profiles, single cell cluster or protein class. This is exemplified by the first-level consensus network for the CCNE1 gene colored according to subcellular location and the CD28 gene colored according to cell type specificity in Figure 1. There is also a highlight option for genes categorised as cell type enriched that will show interaction partners expressed in the same cell type(s) and thus indicate the most probable interactors in the particular cell type(s). Custom highlighting of genes is possible using the query builder in the Filter option. For consensus interactor pairs the 3D structure of the interaction predicted by AlphaFold 3 including predicted template modelling scores and the corresponding predicted aligned error (PAE) plot are also displayed as shown in Figure 2. Predicted interaction between CCNE1 and CDK2
CDK2 - pTM: 0.92
CCNE1 - pTM: 0.69
ipTM: 0.92
View the interaction of CCNE1 and CDK2 in the Human Protein Atlas Predicted interaction between TNF and TNFRSF1B
TNFRSF1A - pTM: 0.35
TNF - pTM: 0.61
ipTM: 0.58
View the interaction of TNF and TNFRSF1B in the Human Protein Atlas
Figure 2. The AlphaFold 3 predicted structures of the interactions between CCNE1 (blue) and CDK2 (red),as well as between TNF (blue) and TNFRSF1B (red) are shown together with the corresponding predicted template modelling scores and PAE plots. Structures are for non-commercial use only, subject to AlphaFold3 Output Terms of Use In the current data 15294 genes have experimentally verified interactions and of these 7298 have at least one consensus interaction, an interaction present in at least two of the datasets. The bar plot in Figure 3 shows the distribution of the number of consensus interactions for these genes. Figure 3. The distribution of the number of consensus interactions. The exact number of genes can be shown using mouse-over. Consensus genes with few interactions seem to a higher degree show tissue enrichment than genes with a higher number of interactions.
Among the 2335 genes with a single consensus interaction 277 are tissue or group enriched, while corresponding number for genes having more than five consensus interactions is 71 out of 2020. The bar plot in Figure 4 shows the fraction of the single cell type specificity categories in all different groups.
Figure 4. Single cell type specificity across the groups with different numbers of consensus interactions. The fraction of genes classified as Cell type enriched is shown in red, Group enriched in orange, Cell type enhanced in purple and Low cell type specificity in dark grey. The number of genes in each category is shown using mouse-over. |