You are correct you do not need DADA2. I believe it would be more useful in your case to utilize the Qiime2 Shotgun Distribution. This workflow should help you work around DADA2 and use Kraken to classify your metagenomic reads. If I am misinterpreting your question feel free to give me more detail so I can better understand your goal. I hope this is helpful and you are able to get your data classified.
But your downstream analysis is failing because you dont have the right file type.
To clarify, are you working with metatranscriptomic data or metagenomic data? I am trying to figure out your goal for this analysis so I can help you with relevant downstream analysis.
If it is metagenomic data you can extract 16S and 18S from your sequences but they will underperform compared to running a metagenomic analysis or having 16S data and running amplicon analysis from there. If you want to proceed in this direction. The commands would be as follows:
qiime feature-classifier extract-reads
But I would again recommend our shotgun distribution for metagenomic data as it is more applicable to your data.
If you have metatranscriptomic data we currently do not have implementation for this type of sequencing.
(1/3) Invalid value for '--i-sequences': Expected an artifact of at least type FeatureData[Sequence]. An artifact of type SampleData[PairedEndSequencesWithQuality] was provided.
(2/3) Missing option '--p-f-primer'.
(3/3) Missing option '--p-r-primer'.
It seems that there are some problems with the command you provided. I will also explore the "Qiime2 Shotgun" approach, but I'd like to resolve this issue since many research papers also use this method.