ENH: allow SampleData[MAGs] as input to predict-genes-prodigal

[Sann5]

What's new

• Input to the predict-genes-prodigal as only allowed to be FeatureData[MAG]. Now it can also be SampleData[MAG]
• Closes ENH: allow gene prediction from SampleData[MAGs] #101

Blocked by...

Preview Give feedback

1. merge after MAINT: adjust regex for genome data semantic types qiime2/q2-types#322

Run it locally

1. Check out the PR branch.

Assuming 1) you already have a local copy of q2-moshpit that is installed in editable mode in your activated virtual environment and 2) the working directory is q2-moshpit; run the following.

PR=154
git fetch origin pull/${PR}/head:pr-${PR}
git checkout pr-${PR}

2. Let's get you some data to play with:

cd my_dir
wget https://polybox.ethz.ch/index.php/s/slGyW1uQbyxmyb9/download -O sample_data_mags.qza

3. Test it out!

qiime moshpit predict-genes-prodigal --i-mags sample_data_mags.qza --o-genes genes.qza --o-loci loci.qza --o-proteins proteins.qza --verbose

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 97.74%. Comparing base (897d7e0) to head (12d4778).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #154      +/-   ##
==========================================
+ Coverage   97.72%   97.74%   +0.01%     
==========================================
  Files          64       64              
  Lines        3390     3415      +25     
==========================================
+ Hits         3313     3338      +25     
  Misses         77       77              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[misialq]

Hey @Sann5, see some refactor suggestions below.

[Sann5]

Hey @misialq 🙌🏼 . I did quite a few changes to the code with your suggestions in mind. Please take a look and tell me what you think. While I have you here I want to clarify something we previously discussed.

This action outputs 3 artifacts whose semantic types are GenomeData[Loci], GenomeData[Genes], and GenomeData[Proteins]. The formats associated with these semantic types have the following structure:

class <name>DirectoryFormat(model.DirectoryFormat):
    <name> = model.FileCollection(r'(.*\_)?<name>[0-9]*\.<extension>$',
                                format=<format>)

• where <name> is either loci, genes or proteins
• <format> is the format of the files like FASTA of gff and <extension> is the associated file extension

We discussed changing the regex to something like

(.*\/)?(.*)\.<extension>$

bascially matching any random file name with an optional parent directory.

Now... for the question... Right now the files that are created in the output artifacts all have this <name> thing in the middle. Should I already remove it (this <name> string)? That would mean that I have to open a PR on q2-types making these changes to the formats and then this PR would depend on that one.

[misialq]

Hey @Sann5, yeah, that would be great! As discussed, there is really no need to have that in the file name - the regex should pretty much be based on the extension itself as the specific formats are anyways linked to their respective semantic types so there is no need to indicate that in file names (it only makes it too restrictive).

[Sann5]

Hey @misialq. I did as we discussed :). I also made the PR as dependent.

[github-actions]

This PR/issue depends on:

• MAINT: adjust regex for genome data semantic types qiime2/q2-types#322

[misialq]

Hey @Sann5, looks good to me! I tried to run through both, sample data and feature data annotation and all worked as expected. Feel free to merge after updating the branch and resolving the conflicts ✅