Hi, Our understanding is that the challenge networks come from different sources and even different providers. Some networks may even be a fusion from several sources. Therefore we can not assume overlapping information between some of the networks on the level of single patients. However, can we assume that the networks represent the same set of diseases? Thanks, Andras

Created by Andras Hartmann ahartmann
Yes, the networks come from different sources. The networks are not based on patient data and they are not disease-specific (except for the cancer network). They are global networks: the protein interaction networks attempt to comprehensively integrate established protein interactions from different databases and the co-expression network is based on a very large compendium of expression data representing a very broad set of conditions. The rationale is that these networks may comprise some functional modules / pathways that could be important for disease processes. We test a very broad set of GWAS traits and diseases. Our initial analysis has mainly focused on individual networks, we haven't yet explored whether some corresponding/overlapping modules from different networks show enrichment for the same set of diseases, if this is what you're asking. (Indeed, we don't know yet if integrating multiple networks for module identification will improve discovery of disease modules - Sub-challenge 2 really asks a question where we don't know yet what the answer will be).

General assumptions about the networks page is loading…