Added more flags to bit get-accs-from-ncbi to make it better mirror its get-accs-from-gtdb counterpart --target-rank added for disambiguation of wanted taxon if needed --get-taxon-counts added --get-rank-counts added --get-table added to turn the parquet-stored data into a tsv for the user Changed bit get-accs-from-ncbi and bit get-accs-from-gtdb now share primarly all the same code for finding accessions based on tax (due to change noted below about ncbi's tables being prebuilt with full lineages baked in) internal changes, not really user-facing bit gen-mg uses less memory when resolving taxonomy the suppression screen now streams the ncbi accession column in batches instead of materializing and caching all of it gtdb lineage/taxid lookups are now restricted to the selected genomes rather than being built for every row of the gtdb table ncbi taxid lookups stream the parquet and stop once all wanted accessions are found the selected-genomes-gtdb-info.tsv lookup is now built only for the selected genomes rather than for every row of the gtdb table the github action that builds the slimmed down ncbi assembly summary data object now builds in full lineages also (so that info is more easily available and always syncd with the cached assembly summary data) gtdb and ncbi data tables are stored as parquets now (smaller/faster downloads, less user storage needed, and are much faster when accessed and searched) more robustness improvements to dl-ncbi-assemblies