Highlights
Two correctness and usability fixes for atb query and atb download.
--sample-fileis fast again.atb query --sample-filenow uses the SQLite index instead of falling back to a full-table scan, so it runs as fast as--samples(a minute-plus → ~0.1s for the reporter's 10 accessions). (#19)- Clearer download vs. data directories. The
--data-dirhelp no longer implies it controls where genomes are downloaded, andatb download --output-dirnow shows its real default path and the config key that changes it. (#20)
Fixed
atb query --sample-filenow uses the fast SQLite index path instead of scanning the whole table. The index lookup read only the--samplesslice and ignored accessions loaded from--sample-file, so a file-based query fell back to a full scan — the reporter's--sample-filerun had not finished after a minute, versus ~0.1s for the same 10 accessions passed via--samples. Both sources are now merged and de-duplicated throughFilters.SampleAccessions()before building thesample_accession IN (...)clause, so--sample-fileand--samplesperform identically. (#19)atbdirectory help no longer conflates the metadata index with genome downloads.--data-dir(and$ATB_DATA_DIR) govern only the local metadata index;atb downloadwrites assemblies to the separatedownload.output_dir(-o/--output-dir). The--data-dirhelp now reads "directory for the local metadata index", andatb download --output-dirshows its effective default path and thedownload.output_dirconfig key. (#20)
Full changelog: v0.17.0...v0.17.1