Discoverability of features

[ivirshup]

Description of feature

How do we make it clear which:

• Annotation columns
• Filters
• Genomes

Are available to be used. GenomicFeatures/ ensembldb do this with a combination of functions for listing available features and pre-defined classes that make these available.

Examples include:

• Available columns: accessor functions like transcripts, axons
• Filters: GeneIdFilter, CoordinateFilter classes
• Genomes: listEnsDbs

Do we copy this design? The filter classes may be quite a lot of effort for our first draft.

---

To implement:

• [x] Filters (gf.filters namespace + doc-site)
• [x] Annotation columns (list_columns, list_tables) (#22)
• [x] Genomes (gf.ensembl.list_ensdb_annotations) (#21)

[emdann]

MVP filter classes IMO would be:

• by gene biotype e.g. get all protein coding
• by ensembl ID
• by gene name
• by chromosome
• by genomic coordinates - any overlap in first instance, no filtering of gene starts/ends, equivalent to GRangesFilter

[ivirshup]

Idea: we can start with a partial set of filters, plus support for ibis expressions. We can then add these in steps.