Proposal - Add --traverse-directory to sourmash compare

[brooksph]

I'm running sourmash compare with a large number of signatures. I'm able to make the comparison using 'sourmash compare *sig' but the same method will not work not work with docker. For example the following command docker run -v ${PWD}:/data quay.io/biocontainers/sourmash:2.0.0a3--py36_0 /data/*sig -k 51 returns

[Errno 2] No such file or directory: '/data/*sig

warning: no signatures loaded at given ksize/molecule type from /data/*sig
loaded 0 signatures total.                                                     
cannot mix scaled signatures with bounded signatures

To confirm that this not an issue with the container I ran docker run -v ${PWD}:/data quay.io/biocontainers/sourmash:2.0.0a3--py36_0 sourmash compare /data/example1.scaled10k.k51.sig /data/example2.sig -k 51 which yielded

0-HKVWJBCXY170605...	[1. 0.]
1-CAVL1ANXX170419...	[0.712 1.   ]
min similarity in matrix: 0.712
loaded 2 signatures total.                                                     
downsampling to scaled value of 1000

Proposal - I think adding '--traverse-directory' to sourmash compare will resolve this issue when run in the following way docker run -v ${PWD}:/data quay.io/biocontainers/sourmash:2.0.0a3--py36_0 /data/ --traverse-directory -k 51 if sourmash knows to look for signatures in the specified directory without providing a pattern.

I started modifying the code but need some help completing the pull request on branch 'add/compare_traverse_directory' right now 'sourmash compare *sig --traverse-directory' yields 'sourmash: error: unrecognized arguments: --traverse-directory' despite changes to commands.py.

[ctb]

that new option sounds good!

two thoughts --

first, the error you're encountering is this, cannot mix scaled signatures with bounded signatures. So somewhere in that directory you have some signatures that weren't computed with --scaled.

second thought - the code on branch add/compare_traverse_directory does add the --traverse-directory option for me so I suspect you are not running your modified code. How are you running sourmash after modifying it?

[olgabot]

should this issue be closed now that the PR #412 is merged?

[luizirber]

thanks @olgabot =]