What is a SObA graph?
Summary of Ontology-based Annotations (SObA)
SObA provides an abridged, summary view of a full ontology graph, trimmed based on annotations and hierarchical relationships.
For inference reasoning, we follow that defined by GO Relations where "is a", "part of", and "regulates" are included in inference.
Starting with the full ontology graph, we edit the graph by:
- (A) removing all nodes (and their edges) that do not carry any (direct or inferred) annotations;
- (B) removing all nodes that do not have direct (as opposed to inferred) annotations unless they are the lowest common ancestral nodes of two or more descending nodes (lowest common ancestor, LCA);
- (C) applying transitive reduction by removing all paths between a pair of nodes unless they are the longest paths;
- (D) reducing hierarchy depth to focus on broader aspects of ontology (top slicing as a user option).
- (E) scaling node sizes based on the relative number of annotations that each node carries to indicate possible relative importance.
Using SObA graph
1. zoom and move nodes around. 2. Select a node or edge for specific information. 3. "Save Image" turns the graph into a high resolution PNG file. 4. Node size scaling may be removed by ticking "Uniform". 5. Displayed graph depth may be set at will (Graph Depth). 6. Graph display may be faceted by evidence types or ontology aspects (GO only).
Types of SObA graph
SObA single gene graph presents the ontology annotations of a single gene
Pair of genes
SObA paired genes combines and compares ontology annotations of a pair of genes.
1. Pick an ontology type. 2. In the text box, enter the name or identifier of the first gene (e.g. lin-3, wbgene00002992, or F36H1.4) and choose from a list of genes returned by auto-complete. 3. Enter and select the second gene. 4. If a gene is not in the list returned, there may be no annotated data.
In the returned SObA graph, each node is sized based on the annotation count for that ontology term, by combining the counts of both genes. A colored slice represents % annotation count. There are two types of slices.
1. Absolute count slices - A slice is % (Annotation count of a gene for node / Annotation count of both genes for node). 2. Relative count slices - A slice is % (Annotation count of a gene for node / Total annotation count for the same gene); a gene slice can take up no more than half of the node area. Thus if there is a single root node (all ontologies but GO), half of it is filled with red and the other half blue, as each gene's annotation count is equal to the total.
List of terms
With the results of an enrichment analysis, one can draw a SObA graph. Enter a list of ontology terms and their corresponding statistical (correct-P or Q) value into the text box. The input may be in the two-column table format. The first column must be terms of the same ontology. The second column is optional.