NEW: filter_kraken2_classifications action

[cherman2]

Hi Bok Lab!

I recently was taking to a collaborator who's area of expertise is metagenomic contaminations/spurious hits. They suggested that I should be preforming low abundance filtering on my data to filter out spurious hits. They suggested that the kraken reports is the best place to filter, so that bracken isn't estimating taxons that were spuriously assigned.

@colinvwood and I develped a methods to allow for filtering the kraken reports by abundance! This method also filters the output files so that the reports and outputs are not out of sync.

Just as a side note: I used this code on a dataset of mine and found that filtering at 0.0001 (which means the taxon has to have roughly 15-60 hits or it gets discarded) retains my diversity signal but minizes the low abundance feature overlap that was happening in my samples!

Let me know if y'all have any questions!

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 95.62%. Comparing base (b6d068a) to head (51a3b7f).
Report is 9 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #226      +/-   ##
==========================================
+ Coverage   95.60%   95.62%   +0.01%     
==========================================
  Files          34       34              
  Lines        1956     2010      +54     
  Branches      226      235       +9     
==========================================
+ Hits         1870     1922      +52     
- Misses         48       49       +1     
- Partials       38       39       +1     

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

Copilot reviewed 3 out of 13 changed files in this pull request and generated 1 comment.

Files not reviewed (10)

• q2_moshpit/kraken2/tests/data/abundance-filter/outputs-only-unclassified/sample1.output.txt: Language not supported
• q2_moshpit/kraken2/tests/data/abundance-filter/outputs-only-unclassified/sample2.output.txt: Language not supported
• q2_moshpit/kraken2/tests/data/abundance-filter/outputs/sample1.output.txt: Language not supported
• q2_moshpit/kraken2/tests/data/abundance-filter/outputs/sample2.output.txt: Language not supported
• q2_moshpit/kraken2/tests/data/abundance-filter/reports-only-unclassified/sample1.report.txt: Language not supported
• q2_moshpit/kraken2/tests/data/abundance-filter/reports-only-unclassified/sample2.report.txt: Language not supported
• q2_moshpit/kraken2/tests/data/abundance-filter/reports-w-unclassified/sample1.report.txt: Language not supported
• q2_moshpit/kraken2/tests/data/abundance-filter/reports-w-unclassified/sample2.report.txt: Language not supported
• q2_moshpit/kraken2/tests/data/abundance-filter/reports/sample1.report.txt: Language not supported
• q2_moshpit/kraken2/tests/data/abundance-filter/reports/sample2.report.txt: Language not supported

Comments suppressed due to low confidence (1)

q2_moshpit/kraken2/classification.py:278

• [nitpick] Consider including the sample_id in the error message for better debugging.

raise ValueError("All Taxonomic bins were filtered by the"

[misialq]

Hey @cherman2, this looks great - thanks! 🏅

Please see the very minor cosmetic comment below.

[misialq]

Should we just put here the text from the "name" field or did you want to keep the ...?

[misialq]

Please provide these missing descriptions 🙏

[nbokulich]

hey @cherman2 @misialq is this PR slightly overlapping with this one?
#212

I think that ideally these actions should be merged into one so that we do not have different kraken report filtering actions. I like that the reports and outputs are filtered in tandem with this action.

[misialq]

Hmmm, yeah, that's a good point actually... I guess the other action focuses on metadata-based filtering but perhaps we could merge them indeed...