Closed krttrr closed 1 year ago
The HPO is a directed graph that points from the leaf terms to the phenotype root term "HP:0000118- Phenotypic Abnormality" which points to "HP:0000001- All."
Therefore the HPO is flipped over in a way- nx.ancestors(G, source)
returns all nodes having a path to the node source
in the graph G
which are the child terms aka farther away from root.
@arvkevi, we could look at the diseases annotated to each node and see how they populate cumulatively from leaf to root. This would give us a good idea of how to answer @krttrr 's question.
Great, I was hoping I was wrong!
Let me know how the other question shakes out...
closing this.
@krttrr the IC is inherited from parents (closer to the root), not from children (closer to leaf nodes), so the answer to your first question is: no.
Related to this, what would the impact be of taking only immediate children rather than all children? That is, are we giving too much IC to less specific terms with this method?