low proportion of merged sequences using the DADA2

Hi all.

I am testing my 16S data using DADA2 on 2x250 paired end reads. This was the quality score box plots.
qual_viz.qzv (309.8 KB)

Based on the results, the sequences were denoised with DADA2.

The script i used:

qiime dada2 denoise-paired --i-demultiplexed-seqs 16s_cb_paired-end-demux.qza --o-table table --o-representative-sequences representative_sequences --p-trunc-len-f 240 --p-trunc-len-r 240 --p-trim-left-f 20 --p-trim-left-r 20 --o-denoising-stats denoising-stats.qza

This was the denoising stats
stats.tsv (3.5 KB)

It looks like most of our sequences were dropped out because they fail to merge.

I wonder why so many reads were lost especially for sample A7, M1, and F1.

Thanks in advance.

Indeed, you are loosing tons reads at merging step. Since there are no issues with filtering step, most likely overlapping region is too short. I would suggest to:

  • set truncation parameter to 0, which will disable it
  • add option to decrease min overlap (check docs how to do it).


Hi, timanix!
Really appreciate for your reply! We will try the methods you suggested.

Best wishes

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.