DADA2: loss 99% of the reads after filtering

Hi all,

I developed a pipeline in QIIME 2 that I use to analyze human samples sequenced with MiSeq (paired-end) and it always works very well. But, with my last set of data almost no sequence passed in the DADA2 filter. The samples were processed in the same way as the others, so, I don't know what is happening.

The version of QIIME 2 I'm running is QIIME2-2019.7 installed by conda.

The command I ran was:
$qiime dada2 denoise-paired --i-demultiplexed-seqs qza/step1_imported.qza --p-trunc-len-f 0 --p-trunc-len-r 0 --o-table qza/step2_denoised_table --o-representative-sequences qza/step2_rep_seqs --o-denoising-stats qza/step2_denoised_stats

These were the results:

Could someone help me to figure this out?

Thank you in advance,

Hi @cvleal, welcome to the forum!

Can you provide us with your demux summary visualization? Also what hypervariable region is this from?
My guess is that you have some some poor quality scores in this run and since you are not truncating any of your low quality sequences nothing is passing the initial filtering in DADA2.
In general, I highly recommend truncating as much of your 3' prime tails as you can (without risking merging issues) in all runs by default, this helps allow more reads to pass the initial filtering and reduces the # of places where spurious reads can slip through. Not truncating anything is generally a bad idea.

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