Figure 1.

Examples on how gene signatures appear in articles, in other databases, and in Liverome. The left, middle, and right columns show gene signatures as appeared in articles, in other gene signature databases, and in Liverome, respectively. Column titles for numerical ranking information were abbreviated or ad hoc in the original gene signature tables (left column). Gene signatures extracted by other databases are uninformative (middle column): (A) direct import of original table causes an ambiguity as to what P1 and P2 means and to the context of the observed expression changes (Down and Up), (B) importing only the gene identifiers causes a complete loss of differential expression information, and (C) importing only the change direction causes a loss of numerical information as well as an ambiguity as to the context of the observed expression changes. Shown in blue highlights how the information from the original signature tables became transformed in the databases. Liverome derives the most informative form of gene signatures through manual curation to construct self-contained database content (right column). In addition, for an easier recognition, fold change values were uniformly formatted (shown in red) and signatures were informatively named (shown in green) in Liverome.

Lee et al. BMC Genomics 2011 12(Suppl 3):S3   doi:10.1186/1471-2164-12-S3-S3