ENH: Add functionality to filter sequence data by Kraken hits

[johnchase]

This is a request for an additional feature, and not a problem with the existing q2-moshpit

Describe the solution you'd like
We would like a function that takes as input sequence data, and the kraken 2 hits (SampleData[Kraken2Output]) and returns two sets of sequence data, one for classified, and one for unclassified

Describe alternatives you've considered
We are currently doing this in pure python

Additional context
We have already started developing this internally, and will use the functionality in our workflows, however, q2-moshpit seems to be a logical place for the functionality

[nbokulich]

Thanks for opening the issue @johnchase ! What type of sequence data do you want to filter?

[johnchase]

For the time being it would be SampleData[SequencesWithQuality], SampleData[JoinedSequencesWithQuality], SampleData[PairedEndSequencesWithQuality], though we could likely add any type of sample data sequences

We're using this to filter contaminants from metagenomic data. In our case the input sequences to classify_kraken2 are the same sequences to be filtered

[johnchase]

@nbokulich I noticed that Kraken2, when run from the command line, has the option to return the classified and unclassified sequences. Would it make sense to include that as default output for the plugin? I can't speak for other users, but the classified and unclassified sequences are the main reason we run Kraken2.