Index all BED files into vector database

[nsheff]

This should be an addition to bedboss --

Each BED file that is added to bedbase will need to have its embedding computed through a model, and then stored in the database.

[nsheff]

• [x] add parameter to bedboss to accept region2vec model (a bbconf config option)
• [x] add config options to bedboss to link to qdrant (bbconf option, like POSTGRES is now)
• [x] use region2vec model to compute embedding of bedfile being processed (use geniml)
• [x] use QdrantBackend object to insert embedding into qdrant

I think it would be great if @ClaudeHu could write the code for computing embedding and inserting into database, and then @khoroshevskyi can actually run it to index the bed files.

[nsheff]

@ClaudeHu can you update on this. Please give @khoroshevskyi the functions he needs to add to bedboss

[khoroshevskyi]

Update: I have added bbconf to accept qdrant and region2vec variables through bbconf config. Changes are on dev branch

[nsheff]

I think the functions you'll need to call in bedboss are almost finished in geniml; we can try to merge these into dev there today and then you can play with that.

@ClaudeHu can you provide a few lines of code here on how to call the new geniml functionality to do this?

[khoroshevskyi]

This functionality already available in bedboss. To index bedfiles, we need first upload them to the bedbase:D

[khoroshevskyi]

Index system is working now. You can reindex all bed files using bedboss