# Problems after using demux on a artifact
**URL:** https://forum.qiime2.org/t/problems-after-using-demux-on-a-artifact/33971
**Category:** User Support
**Tags:** demux
**Created:** [February 9, 2026, 3:55pm UTC](https://forum.qiime2.org/t/problems-after-using-demux-on-a-artifact/33971 "2026-02-09T15:55:39Z")
**Posts on this page:** 7
**Page:** 1
### Author:  [@hgrandes](https://forum.qiime2.org/u/hgrandes)
#### Post date: [February 9, 2026, 3:55pm UTC](https://forum.qiime2.org/t/problems-after-using-demux-on-a-artifact/33971/1 "2026-02-09T15:55:40Z")
# Problems using demux on a artifact
I am using a Windows 11 with ubuntu where I am using Conda for the environment to use the latest version of qiime2 (qiime2-amplicon-2026.1). After creating the artifact from a manifest of two with 39 samples with foward and revers reads sequenced by Ilumina and given in .fastq file. The plot looks almost empty. Why its that?
---
### Author:  [@colinvwood](https://forum.qiime2.org/u/colinvwood)
#### Post date: [February 9, 2026, 6:53pm UTC](https://forum.qiime2.org/t/problems-after-using-demux-on-a-artifact/33971/2 "2026-02-09T18:53:37Z")
Hello @hgrandes,
If you're referring to the quality score plot looking empty this is because what is plotted at each base position is a box plot of the distribution of quality scores. When there is little enough variance among the quality scores at a position the box plot collapses to just a single dash. Not to worry though because the quality scores are high.
Hope this helps,
Colin
---
### Author:  [@hgrandes](https://forum.qiime2.org/u/hgrandes)
#### Post date: [February 9, 2026, 9:05pm UTC](https://forum.qiime2.org/t/problems-after-using-demux-on-a-artifact/33971/3 "2026-02-09T21:05:03Z")
It sees that it has good quality scores but one I performed dada2 denoising it gives me terrible values, filtering practically all the entire sequence of some samples:
For what I understand its that if I want to have a good data I need to have valiues \>75% in the last colim. Or im I wrong and I can use this filtered sequences for further analysis?
---
### Author:  [@colinvwood](https://forum.qiime2.org/u/colinvwood)
#### Post date: [February 9, 2026, 9:28pm UTC](https://forum.qiime2.org/t/problems-after-using-demux-on-a-artifact/33971/5 "2026-02-09T21:28:18Z")
Hello @hgrandes,
You are correct that losing such a large proportion of reads in some samples during the dada2 denoising process is not desirable. However, if those samples are included in the demux visualizer you posted above, then the quality scores are unlikely to be the issue. I would review the truncation parameters that you used to make sure that you do not set either to a value that is longer than a significant proportion of the reads. You can see a summary of the read length distributions in the demux visualizer.
---
### Author:  [@hgrandes](https://forum.qiime2.org/u/hgrandes)
#### Post date: [February 10, 2026, 12:55am UTC](https://forum.qiime2.org/t/problems-after-using-demux-on-a-artifact/33971/6 "2026-02-10T00:55:51Z")
You dont think that may be its a problem with the creation of the artefacts? I think that the format might be wrong. The files that they sent me are .fasq of the foward and reverse reads of each sample. Then I put it into a manifest file in csv format al sequences and then I created the artefact with that. Can you check if my scripts ar ok?
Script of the cvs file:
sample-id,absolute-filepath,direction
G1,PATHWAY,forward
G1,PATHWAY,reverse
G2,PATHWAY,forward
G2,PATHWAY,reverse
…
The creation of the artefacts script:
qiime tools import \
--type 'SampleData[PairedEndSequencesWithQuality]' \
--imput-path PATH/FILE(csv) \
--output-path PATH/ARTIFACT \
--input-format PairedEndFastqManifestPhred33
---
### Author:  [@colinvwood](https://forum.qiime2.org/u/colinvwood)
#### Post date: [February 10, 2026, 4:42pm UTC](https://forum.qiime2.org/t/problems-after-using-demux-on-a-artifact/33971/7 "2026-02-10T16:42:07Z")
Hello @hgrandes,
It also seems unlikely that the import was the problem because the dada2 stats visualization that you posted above shows that the samples have reads (in the "input" column). In the very next column ("filtered") some samples drop almost all their reads. This is why I suggested double checking the truncating parameters, because I think most of what happens between "input" and "filtered" is quality filtering, truncating, and length-based filtering.
---
### Author:  [@system](https://forum.qiime2.org/u/system)
#### Post date: [March 13, 2026, 10:42pm UTC](https://forum.qiime2.org/t/problems-after-using-demux-on-a-artifact/33971/8 "2026-03-13T22:42:49Z")
This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.