Reverting to Qiime2 after using DADA2 in R

[HollyJ]

I have two MiSeq runs of 240 samples each that act as technical replicates. To analyse, I have denoised the runs separately, merged them, and finally removed chimeras all using DADA2 in RStudio (v1.14.0 and v1.2.5033 respectively). I first tried to do this with the dada2 plugin within qiime2 but merging the two runs proved difficult so I switched.

The output from the chimera removal is a single large matrix (8106720 elements, 36.7Mb), and I can’t work out how to save it in a format that I can then upload to qiime2 (v2019.4, linux cluster) for further clustering and taxonomic assignment.

I checked the formats and types that qiime2 can import (although I don’t understand the differences between types and formats, and can’t find documentation explaining them), and I don’t know how I could convert my matrix to fit. Is there a simple .rds to .qzv conversion?

The code I used is just the code in the big data tutorial so I won’t replicate it here.

[Nicholas_Bokulich]

Hi @HollyJ,
There is a community tutorial describing this (not sure if it is out of date):

Importing dada2 and Phyloseq objects to QIIME 2

Importing dada2 and/or Phyloseq objects to QIIME 2 Background This tutorial describes how to take feature/OTU tables, taxonomy tables, and sample data (metadata) from R and import into QIIME 2. This might be useful if you have already completed analyses in R using (but probably not limited to) the dada2 and phyloseq packages and you want to add or compare to data analyzed in QIIME 2. You may also want to use or share data with the nice interactive visualization features of QIIME 2. This will hel…

Good luck!

[HollyJ]

Hi Nicholas,

Thanks, but I’ve already looked at that tutorial and followed the relevant parts. I do not want to assign taxonomy in R, I want to do that step in the pipeline I already have in Qiime2.

That tutorial only shows how to make the feature tables, which I have done, but not the other artifact needed to complete the vsearch cluster step and feature-classifier classify-sklearn steps - the rep-seq part. Do you know how I can do that?

Thanks again,
Holly

[Nicholas_Bokulich]

@HollyJ

which I have done, but not the other artifact needed to complete the vsearch cluster step and feature-classifier classify-sklearn steps - the rep-seq part. Do you know how I can do that?

Down at the bottom of that tutorial:

@ChristianEdwardson

Bonus: Export and Representative Sequences from dada2:

In R:

uniquesToFasta(seqtab, fout='rep-seqs.fna', ids=colnames(seqtab))

In QIIME 2:

qiime tools import \
--input-path rep-seqs.fna \
--type 'FeatureData[Sequence]' \
--output-path rep-seqs.qza

Let me know if that works!

[HollyJ]

Oh… apparently I’m blind! Thank you so much!