Fix KeyError when accessing lineage dictionary in OBO parser

[manvikri22]

Description

This pull request addresses an issue where accessing the lineage dictionary resulted in a KeyError if the superentity was not found.

Changes Made

• Implemented a conditional check to ensure that the superentity exists in the lineage dictionary before accessing it.
• Added a warning message to inform users when a superentity is not found, improving the robustness of the parser.

Motivation

This change enhances error handling in the OBO parser, making it more resilient to inconsistencies in the ontology files and providing clearer feedback to users during debugging.

[althonos]

Hi @manvikri22 , do you have an example where such an error happens? My impression is that the KeyError would happen here if a superentity is not declared for a subentity, so it may be better to resolve this more globally to support dangling entities (#225).

[manvikri22]

Hi @althonos,

Thank you for the feedback!

You're right that the KeyError could occur when a superentity is referenced but not declared (a dangling entity). In the case I encountered, I was working with an OBO file where certain is_a relationships pointed to terms that were either missing or improperly defined. Here's a minimal example that triggers the error:

[Term] id: CL:0000540 name: hematopoietic stem cell is_a: BFO:0000040 ! some undefined biological entity

In this case, the term BFO:0000040 is referenced as a superclass but is not declared in the file, leading to the KeyError when the symmetrize_lineage() function tries to access it. The fix I proposed ensures that the missing entity is handled gracefully by logging a warning, but you're right that this might need a broader resolution to support dangling entities in a more global way.

If you'd like, I can update the pull request or work on a more comprehensive fix based on your guidance.
Do you have specific suggestions or references for the dangling entity resolution? I could also look into updating the parser to log missing entities while preserving the overall structure, or perhaps adding an option to skip undefined superentities gracefully.

Let me know what you think!