The feature_relationship table allows us to store feature graphs. In this model, feature records are nodes and feature_relationship records are edges. For example, consider gene annotations, such as a polypeptide feature, and a computational pipeline that generates a graph of related polypeptides. In this case, we seek to model the graph output of this pipeline in feature_relationship records.
We propose adding a feature_relationship_analysis table to associate the feature_relationship records with the analysis that generated them.