A context-driven gene prioritization method for web-based functional genomics.

Functional genomics experiments often result in large sets of gene centered results associated with biological concepts such as diseases. Prioritization and interpretation of these results involves evaluation of the relevance of genes to various annotations or associated terms and is often executed through the use of prior information in biological databases. These diverse databases are frequently disconnected, or loosely federated data stores. Consequently, assessing the relations among biological entities and constructs, including genes, gene products, diseases, and model organism phenotypes is a challenging task typically requiring manual intervention, and as such only limited information is considered. Extracting and quantifying relations among genes and disease related concepts can be improved through the quantification of the entire contextual similarity of gene representations among the landscape of biological data. We have devised a suitable metric for this analysis which, unlike most similar methods requires no user-defined input parameters. We have demonstrated improved gene prioritization relative to existing metrics and commonly used software systems for gene prioritization. Our approach is implemented as an enhancement to the flexible integrative genomics platform, GeneWeaver.org. Lecture Notes in Computer Science 2013; 7875:161-17