High-throughput technologies are typically used to compare gene expression in different biological conditions. In the Atlas, a biological condition is defined as an experimental variable (or experimental factor, EF). In an experiment, an experimental factor (EF) is the parameter that is analysed to determine changes in gene expression and this EF normally has several values (EFVs). For example, in an experiment that compares gene expression in leukaemic versus healthy blood cells, the EF tested is the 'disease state', which has two values (EFVs), 'leukaemia' and 'normal' (the latter being the ontology term for the healthy state).
For each experiment, we apply a statistical test (6-15) to identify differentially expressed genes for each EF. Following on the 'disease state' (EF) and 'leukaemia' / 'normal' (EFV) example above, for each EFV, a gene is considered 'differentially expressed' if its mean expression under this EFV is significantly different from this gene's mean expression across all EFVs (i.e. across all 'leukaemia' and 'normal' samples).
Finally we aggregate all the identified gene-EFV-differential expression associations for storage and retrieval (Figure 2).
The result of aggregating all gene-EFV-differential expression associations is the Atlas itself which can be queried through its interface.
Figure 2. The Atlas can be considered as a data matrix, where the matrix entry is the information on the experiments in which the gene was found to be differentially expressed, the columns are biological conditions and the rows are genes.