PDS Project: Added text2term Ontology Mapper to ChatGSE

[RM7]

As part of our PDS project, we developed an ontology mapper with the help of text2term and added it to ChatGSE.

[slobentanzer]

Thanks a lot for the PR, great work! There are some adjustments that need to be made for it to be publishable, but that should be fairly straightforward. I will collect improvements to be made here (that does not mean that you are responsible for implementing them all).

On first glance:

• the ontology mapper and cell type annotation tabs are redundant
• the ontology mapper module should be moved to biochatter
• the mappings should not be written to some file by default, they should stay in memory and formatted in JSON if the user wants to download them
• the threshold of automating annotations (trust) should be programmatically available to the user
• the mappings should include CL IDs, not just plain names
• extracted columns should be up to the user (I don't think there is a consensus on what to call cell type columns); alternatively, no column names at all should also be an option
• mapping by LLMs and assistance to text2term by LLMs should be made possible to supplement in edge cases
• visualisation of clusters and embeddings can help support the annotation decision
• we should think about a way to include cell type markers in the workflow
• include cell states (functional annotation)? @roramirezf @JanLanzer