It's always a shame when a run does not sequence well
Overloading the Illumina flow cell. Underloading the flow cell. Poor amplification during PCR, and everything that can cause that.
Hopefully it's a sequencing issue and not a library amplification issue. I would ask if the sequencing core would be willing to resequence these samples, hopefully for free or maybe a discounted price.
Sure! I'm worried about the quality drop at base 20 in the forward read, but it's always worth a try...
In addition to the possible reasons already mentioned, the beginning of the forward reads makes me think that the sequencing library might not have had enough nucleotide diversity in some positions. When all bases are the same in a given sequencing cycle, the instrument software has trouble confidently calling bases (more info on this topic here). These could be primers at the start of your amplicon or conserved regions of the target gene. To avoid this, we usually add at least 5% PhiX or some library with a random distribution of nucleotides (a genomic library of a bacterial isolate, for example) in our MiSeq runs of metabarcode sequencing. If you arrange resequencing of these samples, it might be a good idea to check whether the nucleotide diversity issue is properly addressed in the sequencing setup.
Thanks for your help and the explanation! I imagined something like that...
Unfortunately, we will not be able to redo the analyses.
I tried using FLASH and Prinseq to do some cutting and joining of the forward and reverse , but I'm getting a very low value of input numeric (dada_stat). I will try not joined the forward and reverse and do a 'no aggressive' denoising.