The Gene Expression Atlas  (the 'Atlas') database provides information on gene expression patterns within different biological conditions. It allows you to search for gene expression changes measured in various cell types, organism parts, developmental stages, disease states, and many other biological/experimental conditions.
- contain all the information needed to properly describe them (according to the MIAME (10 ) or MINSEQE  guidelines);
- contain a large number of samples and enough sample replication to increase the robustness of the statistical analysis;
- contain all the data files required for the analysis (6 ).
Once a dataset is selected for the Atlas, its sample annotation is improved through the use of Ontology terms that enables you to formulate searches using biological concepts. The Experimental Factor Ontology (EFO) has been developed to support complex queries across EBI functional genomic resources (Archive and Atlas). As an example, if you use the term 'cancer' to query the Atlas (Figure 1), your search will include all the biolgical process, disease states, different organisms and the different biological sites (colon, breast and others) (A) in which the term cancer was found. The result of your search will be then diplayed as node of a tree (B) in the Atlas.
Figure 1. The EFO brings together all the terms used to describe experimental conditions in the Archive/Atlas experiments, and connects them in a logical way.
Gene expression is measured on a microarray by DNA probes (or oligonucleotides) designed to correspond to part of a known gene, known transcript or putative open reading frame (ORF). For proper interpretation of microarray data, it is important that each probe is annotated with information such as its sequence and corresponding genome location. To this end, microarray  annotation for different chips/platforms in the Atlas is regularly updated to the latest genome assemblies and gene annotations, using either Ensembl  or Uniprot . Updating an array annotation invovles re-mapping all probes on the microarray to the latest version of the reference genome and gene annotation to make sure that no probe is wrongly annotated, e.g. probes mapping to a portion of the genome that ceases to exist in the latest assembly due to the correction of previous sequencing errors, or those mapping to a genomic location that no longer contains an ORF/gene because of a change in either genome assembly and/or gene annotation. If a probe is wrongly annotated, it will be discarded from any subsequent analysis.