On entropy and information in gene interaction networks.

摘要: Motivation Modern biological experiments often produce candidate lists of genes presumably related to the studied phenotype. One can ask if gene list as a whole makes sense in context existing knowledge: Are reasonably each other or do they look like random assembly? There are also situations when one wants know two more sets closely related. Gene enrichment tests based on counting number have common adequate we presume that only fact identical. If by mean well connected interaction network space, need new measure relatedness for sets. Results We derive entropy, information and mutual networks, starting from simple phenomenological model living cell. Formally, describes set interacting linear harmonic oscillators thermal equilibrium. Because energy function is quadratic form degrees freedom, entropy all derived quantities be calculated exactly. apply these concepts estimate probability several independent genome-wide association studies not mutually informative; disjoint canonical metabolic pathways infer relationships among human diseases their signatures. show present approach able predict observationally validated detectable methods. The converse true; methods therefore complementary. Availability implementation functions defined this paper available an R package, gsia, download at https://github.com/ucsd-ccbb/gsia.