DADA2--data preparation

[eDNA]

In this poster (Denoise-single error associated with --p-trunc-len values), @benjjneb recommended importing non-merged reads. Would there be any differences between importing single reads and merged reads if the read-merging program can handle the quality for overlapping bases and update base qualities?

There are a few ways that I can prepare the data before using QIIME2:
1 Cut adapters, trim poor quality bases, trim Ns, merge reads;
demultiplex merged reads;
import as single-ended reads.

2 Cut adapters, trim poor quality bases, trim Ns;
Demultiplex non-merged reads
import as pair-ended reads.

3 Cut adapters;
Demultiplex non-merged reads
import as pair-ended reads

Do you have any suggestion about which one may be the best for using DADA2?

We used dual barcodes in both primers, thus demultiplexing merged reads may be more accurate to assign reads to samples.Thanks.

[benjjneb (Ben Callahan)]

We recommend 3, provided you can accurately demultiplex that way.

You can try running the merged reads, we simply haven’t tested that on your particular read-merging program, so we cannot say whether it will work well or not. It might be fine. It might not, if the qualities assigned by the read-merging program aren’t consistent with the unmerged quality scores, which is the case with many read-merging programs.

[eDNA]

Thank you for your help.