Use as a Python library

[SandyRogers]

This PR:

• refactors the code a bit so that steps 1-3 can be done from a Python script, i.e. use assembly_uploader as a Python library
• adds a test suite for the python library usage
• tweaks the assembly study title/abstract generation (just formatting issues)
• clarifies a few other points in the README

In short it makes this possible:

from assembly_uploader.study_xmls import StudyXMLGenerator, METAGENOME
from assembly_uploader.submit_study import submit_study
from assembly_uploader.assembly_manifest import AssemblyManifestGenerator

StudyXMLGenerator(
    study="SRP272267",
    center_name="EMG",
    library=METAGENOME,
    tpa=True,
    output_dir=Path("my-study"),
).write()

new_study_accession = submit_study("SRP272267", is_test=True, directory=Path("my-study"))

AssemblyManifestGenerator(
    study="SRP272267",
    assembly_study=new_study_accession,
    assemblies_csv=Path("/path/to/my/assemblies.csv"),
    output_dir=Path("my-study"),
).write()

[Vkale1]

Thank you very much ⭐

[Vkale1]

Is it okay to assert in a non test script?

[SandyRogers]

Good q – sort of. I think it makes sense here because: in CLI mode, the user input of library is already validated because I added choices= to the argparse. (Generally you should not use assertions to check user input, because assertion checks can be disabled at runtime by running python in "optimise" mode.) So here the assertion can only fail through programming error – where assertions are useful to sanity check things that should always be true. Would be slightly better if the "library user" input was validated differently (e.g. a Pydantic type or explicit check) but in my opinion this kind of check is okay for that, and the stacktrace for this failing should be self explanatory.

[Vkale1]

'dropbox'? and below

[SandyRogers]

Thkna yuo!