I work on the reconstruction of 16S and 18S sequences from metagenome data (specifically enriched in rDNA). To this end, I am used to apply EMIRGE (Miller et al., 2011) and MATAM (Pericard et al., 2018) softwares on my sequencing data to get near full-length rDNA sequences. The 2 softwares ouput sequences from 800 and ~1500 pb for 16S. Consequently, I get sequences of different length but having taxonomic assignation potentiality far more precise than amplicons, and so the best will be to keep all sequences.
I am wondering in this case if Dada2 (or Deblur) could be suitable to highlight sequence variants?
Many thanks to your reply and comment!
I can't speak super authoritatively here, but it sounds like you don't actually need DADA2/Deblur as you already have your sequence variants from your current pipeline.
So I would probably skip straight to the feature-table step and use your sequences directly. You'll probably need to use qiime vsearch dereplicate-sequences to get your table, but even that may be unnecessary.
thank you very much for your quick reply.
I was supecting that the DADA2/Deblur step was not necessary and here you consolidate my opinion.
Data outputed from my pipeline contain reconstructed rDNA sequences (unique sequences) and their relative abundance among all reconstructed rDNA sequences (annotated in the fasta header). So I guess that I just need to pick relative abundances and calculate the corresponding number of reads before further processing.