Can prejoined sequences be used for Dada2 pipeline

Hi,
I am a beginner for Qiime2 and after reading the tutorials I started analyzing my data. I got OTU table after applying Dada2 pipeline and I noticed that merging for most of the samples have failed . As a result I lost most of the samples from otu table. Could you please provide your expert opinion on improving the parameters for getting all samples.
I used following commnads
qiime dada2 denoise-paired --i-demultiplexed-seqs demux.qza --p-trim-left-f 25 --p-trim-left-r 20 --p-trunc-len-f 300 --p-trunc-len-r 300 --p-max-ee-f 2 --p-max-ee-r 2 --p-trunc-q 2 --p-chimera-method consensus --p-min-fold-parent-over-abundance 1 --p-n-threads 1 --p-n-reads-learn 1000000 --p-hashed-feature-ids True --o-table table.qza --o-representative-sequences rep-seqs.qza --o-denoising-stats denoising-stats.qza

Another point is that I also have demultiplexed and mergepair reads for each sample as provided by the sequencing company. Can I use these prejoined and demultiplexed fastq files for Dada2 pipeline. If yes, then

  1. what parameters I shall consider in Dada2 pipeline so as to avoid merging?
  2. For importing fastq manifest files of these in qiime 2 shall I use “input-format SingleEndFastqManifestPhred33”?
    Thanks for your help.
1 Like

Hi!
I used premerged paired reads in Qiime2 once and imported them as single reads

Hi, many thanks for your reply. Did you use Dada 2 pipeline after importing? If yes, could you please tell me the detail. Thanks and regards

You cannot use DADA2 on premerged or pre-quality filtered reads. If you wish to do merging or quality filtering before hand, Deblur is a good denoising option.

It is also possible and valid to simply use the single end reads without jointing (you retain only the forward reads) in cases where reads aren’t long enough to merge or the reverse reads are too low quality.

Best,
Justine

2 Likes

For importing the merge pair end reads I used following command:
qiime tools import
–input-path paired_manifes.csv
–output-path demux.qza
–type SampleData[JoinedSequencesWithQuality]
–input-format SingleEndFastqManifestPhred33

Hi,
Thank you so much. I tried deblur now. Since deblur uses two steps so I used first step using following command:
qiime quality-filter q-score** --i-demux demux-qza --o-filtered-sequences demux-filtered.qza --o-filter-stats demux-filter-stats.qza
and after this I got following error:
(1/1) Invalid value for “–i-demux”: Expected an artifact of at least type
SampleData[SequencesWithQuality | PairedEndSequencesWithQuality]. An
artifact of type SampleData[JoinedSequencesWithQuality] was provided.
Any idea how to deal with this error?
Thanks and regards

HI @sanda,

What version of QIIME 2 are you running?

Best,
Justine

I am using qiime2-2019.7

Hi @sanda,

In that case, I would try importing your reads as SampleData[SequencesWithQuality], because it should be the closest solution.

Although it’s odd, because I know this works on sequences joined in vsearch, which produces that semantic type.

Best,
Justine

Hi @jwdebelius
Thanks. You mean I should try to import sequences again. Before I used following command to import merged reads
For importing the merge pair end reads I used following command:
qiime tools import
–input-path paired_manifes.csv
–output-path demux.qza
–type SampleData[JoinedSequencesWithQuality]
–input-format SingleEndFastqManifestPhred33

So is this wrong?

Hi @sanda,

I would use the same command, but change your semantic type (--type) from SampleData[JoinedSequencesWithQuality] to SampleData[SequencesWithQuality].

It isn’t that anything is wrong, persay, just that sometimes when things don’t behave quite the way you expect, you have to try things that will probably work and troubleshoot.

Best,
Justine

Hi @jwdebelius,
Thanks for your suggestion. Ok I will try accordingly.

Best Regards

1 Like

Hi @jwdebelius,
After importing merged sequence reads as SampleData[SequencesWithQuality] now I got quality plot. I want your suggestion whether this quality plot is OK or not. And for p-trim-length which value is OK to be used for Deblur pipeline. I am attaching quality plot for your kind reference.
Thanks and Regards

image

1 Like

Hi @sanda,

I think your quality plot looks fine. There’s no magic number for where to trim, although I try to keep like 80% of my sequences at that trim length.

Best,
Justine

3 Likes

Hi,
I am just wondering about the filtering steps. After denoising steps (Dada2 or Deblur), do we need to filter again using another plugin such as De novo chimera filtering with vsearch or borderline chimera removal steps. Because in my case, after denoising step, I am getting huge feature table and I want to cut down the number of OTUs for further analysis. Is this logical or not?
Thanks

Hi @sanda,

Both DADA2 and Deblur apply chimera filtering bu default. Dada2 has a chimera filtering step, and Deblur filters against the greengenes database to remove putative chimeras, mitochondrial, chlorophasts, etc.

You should keep your full table for diversity analyses if you plan to use traditional, rarefaction-based metrics (like in the moving pictures tutorial). Im not as certain about filtering with things like DECOIDE and similar, so I would check the individual papers/documentation.

I think you will need to filter during the feature-based analysis step. However, there’s still discussion about what kind of filtering is best. You could choose relative abundance, relative prevalence, variance based, some composite…

Best,
Justine

1 Like

@jwdebelius,
Thank you so much for your sincere advice.

Best regards,

1 Like

A post was split to a new topic: Rarefying and taxonomy