It appears that reads wrongly orientated were going through the DADA2 pipeline, causing spurious alignements in phylogenetic tree, which could cause distortion in Unifrac distances for instance.
The best solution would be to reverse complement the reads if needed. Yet cutadapt current version of cutadapt 2.10 is not capable to do that for paired-end read.
This is not a problem with vsearch since PANDAseq removes sequences not presenting the forward primer on forward reads and the reverse primer on reverse reads.
It appears that reads wrongly orientated were going through the DADA2 pipeline, causing spurious alignements in phylogenetic tree, which could cause distortion in Unifrac distances for instance.
The best solution would be to reverse complement the reads if needed. Yet cutadapt current version of cutadapt 2.10 is not capable to do that for paired-end read.
This is not a problem with vsearch since PANDAseq removes sequences not presenting the forward primer on forward reads and the reverse primer on reverse reads.