Table 1.

Seven ways to boost power with knowledge

Power Boost	Example Use in Filters or Integrators
Sequences & structures	Filtering GWAS to focus on SNVs close to gene coding regions (Segre et al., 2010). Integrating information about multiple SNVs that fall near the same gene (Segre et al., 2010).
Molecular function (Gene sets)	Integrating individual signals from functionally related genes (Chasman, 2008; Wang et al., 2010).
Molecular networks (Interactions)	Projecting data on networks to identify novel pathways enriched for differentially-expressed genes or candidate SNVs (Ideker et al., 2002). Filters can prioritize gene candidates by their network proximity to known disease genes (Lage et al., 2007) or via signaling pathways that connect gene knockouts to their downstream effects (Suthram et al., 2008; Zhu et al., 2008).
Multiple layers of data	Filtering based on multiple layers of data can retain true signals which are reflected coherently across data sets and filter out noise which usually cancels out (Lee et al., 2004).
Evolutionary conservation	Evolutionary filters retain information that is preserved across multiple species and therefore likely represents functionally relevant signal (Erwin and Davidson, 2009; McGary et al., 2010).
Inherent modularity	Inherent biological modules identified by clustering or eigenvalue decomposition can be used to integrate the stream of information on individual biological entities to a new stream of information about modules (Zhu et al., 2008).
Focusing phenotype by 'Asking the right question'	Phenotypes of interest are often confounded by miscellaneous factors such as age, gender, race, or geographic location. Subtracting away these factors represents a powerful filter that reduces information irrelevant to the analysis (Segre et al., 2010).