Figure 2

Database relational table schema for MatchMiner. (a) Logical database representation. Data are incorporated from the UCSC Human Genome Build, LocusLink, UniGene, OMIM, and the Affymetrix annotation sets for HU95 and HU133 chips. Each candidate gene is assigned a gene index in the GeneIdx table. These gene indexes are used as keys for all of the MatchMiner operations. The number of many-to-many relationships in the model illustrates the complexity of the data. (b) Physical representation of the database. The implementation currently includes 14 tables with about 12 million rows.