Skip to main content
Figure 1 | BMC Bioinformatics

Figure 1

From: Graph-based iterative Group Analysis enhances microarray interpretation

Figure 1

Principle of Graph-based Iterative Group Analysis. A Evidence network. Genes are associated with their annotation in the form of a bigraph (two types of nodes). B The same evidence represented as a simple network. Genes that share an annotation are connected. C-H Example of a GiGA analysis using fictitious microarray results. C Genes are assigned their ranks based on observed expression changes. D Local minima are found, i.e. genes that have no connection to genes with a better rank. E-H Iterative expansion of subgraphs from one of the local minima, gene 2 (rank 1). E The neighboring node with the smallest rank is included (gene 4, rank 4), which leads to the additional inclusion of genes 5 (rank 3) and 6 (rank 2). F Gene 3 (rank 5) is included). G Gene 7 (rank 7) is included, leading to the inclusion of gene 8 (rank 6). H The last gene reachable from this local minimum, gene 1 (rank 8), is included and the process terminates. For each of the subgraphs a p-value can be calculated (see text) and the subgraph with the smallest p-value is declared the "regulated neighborhood" of the local minimum. In the example, genes 2, 4, 5, and 6 form a regulated neighborhood (p = 0.014). The graph expansion process would then be repeated for the remaining two local minima.

Back to article page