'Too few arguments' Error while trying to download genomes with specific taxIDs

Hello, Thanks for making bulk genome download from NCBI so much easier! I am looking to download multiple genomes of the species Mycobacterium tuberculosis (taxid: 1773) from refseq/genbank so that I can construct a pangenome with Roary (https://github.com/sanger-pathogens/Roary).

I have looked into the following issues #66, #65, #55, and #14 for some clarification but remain confused as to why I see the too few arguments error when I run ncbi-genome-download with the following options: ncbi-genome-download -s 'refseq' -F 'gff' -l 'complete' -t "1773,83332" -R 'all' -o mtb-genomes/ -H -n -m METADATA_TABLE -v

I have tried various combinations of this & with individual taxIDs too (with -t and -T). I have tried with the default output folder, without the -v, -H, -n (test), -m, options. I always see the following error:

usage: ncbi-genome-download [-h] [-s {refseq,genbank}] [-F FILE_FORMAT]
                            [-l ASSEMBLY_LEVEL] [-g GENUS] [-T SPECIES_TAXID]
                            [-t TAXID] [-A ASSEMBLY_ACCESSIONS]
                            [-R {all,reference,representative}] [-o OUTPUT]
                            [-H] [-u URI] [-p N] [-r N] [-m METADATA_TABLE]
                            [-n] [-N] [-v] [-d] [-V]
                            group
ncbi-genome-download: error: too few arguments

Environment: macOS mojave. ncbi-genome-download version 0.2.7 downloaded with updated pip.

Please advise. Thank you!

[jrjhealey]

You’re missing the positional argument of what database type to use.

Just append bacteria to the end of your existing command.

[jrjhealey]

I’m not at a terminal to test super thoroughly at the moment, so I’m not 100% sure what the distinction between -t and —genus will result in. The genus option is a basic string match though IIRC, so it might be that there are accessions assigned to that taxid that do not have your particular genus string in their metadata names. This is a guess though, @kblin would probably have to weigh in.

As for gimme_taxa, I’m afraid it doesn’t offer any such filtering. The idea behind the script originally was to get an exhaustive list of tax ids to pass in to ngd which could then use all the normal filters to just ignore taxids that didn’t meet the criteria. Your complete filter should return the same end result though given an equivalent set of input taxids (I would expect).