Human protein-protein interactome: IMEx's metadata on host tissue specificity and genetic variability of proteins

Molecular interaction networks provide maps to explore cellular processes from a systems perspective. Access to the experimental evidence and contextual metadata which influence the interaction outcome are critical for accurate interpretation of interaction dynamics.

The IMEx consortium (www.imexconsortium.org) is an international collaboration that compiles molecular interaction data in full experimental detail from scientific literature and direct submissions. This information is freely available to the public through an open source, open data model.

Each interaction receives an IMEx-level confidence score based on interaction type, detection methods, and the publications reporting the interaction. Factors affecting interaction strength are annotated at experimental, interaction, and participant levels, including clinical and experimental mutations, required or sufficient binding regions, post-translational modifications (PTMs), and external modulators such as agonists and antagonists, derived from low-throughput experimental evidence.

Our curation strategy allows us to map interactions to the specific host tissue or cell type where they were detected. Recently, we completed mapping our host tissues and cell types to the Experimental Factor Ontology (EFO). Modern high-throughput methods like affinity purification-mass spectrometry (AP-MS) can generate protein-protein interactions (PPIs) in their close-to native state of expression.

IMEx contains over 11,800 n-ary interactions, mostly AP-MS evidence captured in human neurons, including more than 3,000 interactions affected by mutations. Approximately 2,200 interactions have been detected in embryonic stem cells. In the bronchial epithelium, more than 3,800 interactions have been detected, with 3,500 of these involving mutation data. Additionally, 500 interactions have been identified in bronchial epithelial cells with a cystic fibrosis phenotype. Close to 500 interactions have been curated in human retinal tissue, over 500 in blood immune cells, and around 350 in human blood plasma and serum. Most references feature PPI data studied in cancer cell lines: over 8,500 interactions have been curated in colorectal cancer cell lines, 6,000 in human breast cancer cell lines, and more than 4,400 in human glioblastoma cell lines and tissue samples. We also capture proximity assays to expand our coverage and provide data on proximal protein partners in different cellular compartments.

IMEx"s comprehensive approach to building metadata for each interaction, with consistent use of controlled vocabularies and ontologies, can generate interaction networks specific to various layers of information. This facilitates the detection of macromolecular assemblies under different conditions.