DADA2 multiple sequencing runs - de novo picking allows for merging after artifact?

[ben (benjamin w.)]

To whom this concerns,

Been forging ahead regarding the use of QIIME2. One question, regarding merging runs that doesn't seem to be answered quite clearly.

In our equivalent qiime1 merge, we would merge after the merging of joined_paired_ends.py (once we get the sequence.fna). But one thing I remember doing is having to specify the sequence # and for different MiSeq runs starting that number differently, in order not to overlap for picking.

In QIIME2, that merge step, knowing the effects of batching on MiSeq runs, is done after the dada2 step. The merging is likely due to the fact that everything is de novo. My question is:

1. If we wanted to re-do with UCLUST tree-based closed or open picking, the merging of MiSeq files should be done prior to this step? If so, is there a guide? Or is it the same way we merge in QIIME1? (in other words, use QIIME1 to perform this).

2. Creation of a tree? The phyloqenetic tree is treated after the DADA2 step, however if we decide on a denote tree creation, is there a way that the tree can be matched with a known tree created from something like Greengenes or SILVA?

Not quite sure if this was answered in a different method, thank you for your help!

Ben

Prior threads/tutorials I referenced:

DADA2- Samples from Multiple MiSeq Sequencing Runs

I am working on implementing a QIIME2 pipeline using the DADA2 plugin. Since ours is a high-throughput sequencing facility, we have 1000s of samples in one project and these samples come from multiple MiSeq Runs. Usually for our Mothur-based Pipeline, we put the FastQs for these samples from separate sequencing runs together using the Manifest File provided and then run them through the pipeline. I am not certain if I can utilize the same process with QIIME2/DADA2-plugin as well? (Means if it’…

https://docs.qiime2.org/2017.8/tutorials/fmt/

[Nicholas_Bokulich]

@ben

If we wanted to re-do with UCLUST tree-based closed or open picking, the merging of MiSeq files should be done prior to this step?

correct, if you use dada2 or deblur, merge the feature table and reference sequences after denoising.

for OTU picking, merging sequences is easiest before OTU picking but can be done afterwards (that is one advantage of closed-reference OTU picking, actually).

open-reference OTU picking can also be done sequentially (e.g., if you have an old dataset that you want to compare to a new one). cluster-features-open-reference outputs a new-reference-sequences file, which contains the reference sequences plus the new de novo OTUs. Use this file as the reference-sequences for the next set of sequences. You can then merge those feature tables and sequences together.

@ben

• If so, is there a guide? Or is it the same way we merge in QIIME1? (in other words, use QIIME1 to perform this).

We have a guide on otu picking but not for merging multiple runs, as I described above. You can also use qiime1 (or method of choice) to merge your sequences prior to importing to QIIME 2 if that is your preference.

@ben

Creation of a tree? The phyloqenetic tree is treated after the DADA2 step, however if we decide on a denote tree creation, is there a way that the tree can be matched with a known tree created from something like Greengenes or SILVA?

sounds like you are looking for q2-fragment-insertion. Head over there and see the tutorial linked from that post for more details.

I hope that helps!