Skip to main content
. 2009 May 25;37(Web Server issue):W147–W152. doi: 10.1093/nar/gkp384

Figure 1.

Figure 1.

Neural network model for the PosMed gene search algorithm. As an example, the user's keyword ‘diabetes’ can be found in several documents, including MEDLINE (Input). These documents are mapped to genes that are supported by manual curation (Concept). Using biological knowledge (e.g. protein–protein interaction, co-expression and co-citation of document sets), PosMed can also suggest genes that do not have the user's keyword ‘diabetes’ in their associated documents (Association). PosMed then returns the candidate genes that are located within the user's specified genomic interval (Output). Thereafter, the user's keyword will be highlighted in the documents (Display).