Our proprietary human interactome network was built by consolidating experimental data from twenty-two unique sources. It contains experimentally derived physical interactions using low and high throughput methodologies.
To further enhance the network and introduce weighted high confidence interactions, relative gene expression, and co-expression data are incorporated to assess the interaction confidence based on the presence of two interacting partners. High throughput perturbation and response data are combined to evolve the interactome from binary to regulatory-directed interactions. Downstream expression patterns of different cell types are used to provide cell specificity.
Using our interactome and network science and graph theory-based algorithms, we reduce the massive amount of features available in large gene and protein expression datasets down to what is biologically relevant to derive individual patient disease signatures.