More interestingly, they used this network

to identify, test and validate novel therapies. Their final networks consisted of a high-confidence set of experimental data points as well as gene targets not included in the original set, but rather added through known protein-protein interactions. From this network set, they systematically expanded targets for therapeutic intervention by identifying targets with known chemical inhibitors and ranking them based on their proximity to the core functional network. From this target set, they identified compounds in clinical trial with known effects on cancer systems and chemical inhibitors not yet tested for GBM [29]. In the interferon-stimulatory DNA (ISD) sensing pathway, an integrated network approach proved successful for identifying novel regulators of this process INCB018424 purchase and for testing new therapeutics [30]. In this analysis, the authors created an interaction network of potential ISD regulators by combining direct interacting partners of known ISD pathway components with interacting pairs from their own quantitative

mass-spectrometry experiments. Perturbation of this compendium network with RNAi reagents Imatinib order identified Abcf1, Cdc37, ad Ptpn1 as effectors of the ISD-sensing response to dsDNA. In this situation, curating and expanding interaction information around known pathway components successfully identified novel genes for the ISD response. The authors also measured ISD-pathway induction after treatment with chemical inhibitors against their novel genes and demonstrated a reduction in deleterious

interferon production. These results show that integration is useful for developing new hypotheses for therapeutic development and supports the Jones et al. perspective concerning efficacy of designing therapeutic options around downstream pathway physiology [26]. Data integration within a network framework also added depth to understanding metabolic disorders using SNP and genetic linkage new data [31]. In this investigation, researchers created a network where interactions depended upon significant co-expression and linkage data between genes. Using optimization, they selected highly connected gene sub-modules and then used these modules for further hypothesis generation. Many sub-modules were enriched for genetic features that were significantly associated with disease traits (fat mass, weight, plasma insulin levels, etc.) and one sub-module was significantly enriched for genetic features with significant correlation to all disease traits. They expanded this module, and created a macrophage-driven superior module from which they selected and further perturbed genetic loci. From these perturbations, they were able to demonstrate the sub-network’s contribution to the observed disease traits and classify genetic features previously not associated with metabolic traits.