adjust Python API for loading and saving signatures?

sourmash-bio / sourmash

Quickly search, compare, and analyze genomic and metagenomic data sets.

http://sourmash.readthedocs.io/en/latest/

Other

473 stars 80 forks source link

adjust Python API for loading and saving signatures? #1312

Open ctb opened 3 years ago

ctb commented 3 years ago

Some ideas rescued from https://github.com/dib-lab/sourmash/issues/1142 -

load_file_as_signatures doesn't take a string input, per https://github.com/dib-lab/sourmash/pull/1304. This challenges it as a replacement for sourmash.signature.load_signatures... I wonder if we can let load_file_as_signatures take a fp, instead of a string as load_signatures does? Then we could support file objects and (via io.StringIO) string inputs.
do we we want to change save_signatures to require a file output, OR alternatively provide a new top-level function that requires you write to a file?

ctb commented 3 years ago

side note, save_signatures should maybe be adjusted to require kwargs, something like

def save_signatures(sigs, fp=None, *, compression=0)

ctb commented 3 years ago

https://github.com/dib-lab/sourmash/pull/1493 provides flexible signature saving functionality that supports Zip files, directories, and sig files.

We might want to add something to this class that makes it a single function call to write a bunch of signatures. Right now it requires multiple calls b/c of the primary use case, but this is easy to fix.

Other changes might include using ZipStorage underneath https://github.com/dib-lab/sourmash/issues/1441.

ctb commented 3 years ago

(the "single function call to write a bunch of signatures" is a good next issue!)

ctb commented 2 years ago

ref https://github.com/sourmash-bio/sourmash/issues/1877