Database path should not default to gtf path

[edurand]

Hello,

Thanks for the great work on pyensembl - very useful. My case might not reflect the general usecase, but I'd like to hear your thoughts about this.

I am creating a Genome instance from local genome files (gtf, fasta, ...). Importantly, I do not have write access to where the genome files are stored.

Then, when I try to index the new genome object, pyensembl attempts to create a database in the same directory as the gtf file, which causes a permission error.

My current work-around

data = Genome(reference_name=build,
                  annotation_name=name,
                  annotation_version=version,
                  gtf_path_or_url=gtf,
                  transcript_fasta_paths_or_urls=fasta)
# Without this I cannot index
data._db.cache_directory_path = data.download_cache.cache_directory_path
data.index()

I know that I can specify cache_directory_path when building the Genome() object, but then I have to re-implement the nice directory structure logic already implemented in DownloadCache, which I'd like to avoid.
Wouldn't it be preferable to have data._db.cache_directory_path default to data.download_cache.cache_directory_path?

Thanks!

[iskandr]

Shipped in v2.6.13 (PR #340).

When a Genome is built from a local GTF without an explicit cache_directory_path, the sqlite Database (and protein/transcript SequenceData pickle caches) now write into the resolved download-cache directory instead of next to the source GTF — so your read-only-GTF-directory case works out of the box. The workaround you posted is no longer needed:

# Before: needed manual fix-up after constructing the Genome
data._db.cache_directory_path = data.download_cache.cache_directory_path

# Now: just construct and index — sqlite and FASTA pickles land in the cache
data = Genome(reference_name=..., annotation_name=..., gtf_path_or_url=gtf)
data.index()

Database(...)'s standalone behavior (write next to the GTF when no cache_directory_path is passed) is unchanged, so direct callers aren't affected. pip install -U pyensembl to pick it up.