Can you generate ASVs from pre-merged paired-end reads?
1
0
Entering edit mode
7 months ago

Title says it all. I'm generating ASVs from several studies' datasets, but some have reads were merged prior to uploading to NCBI or wherever.

My PI tells me that dada2 can't handle this sort of thing (I haven't run through it myself, but I gather that it might obscure the true biodiversity even if dada2 doesn't give a flat-out error). We were going with mothur & vsearch to make OTUs earlier in the process, but mothur doesn't play nice with snakemake and dada2 would be better anyway.

Anyone have any pointers? Help would be very much appreciated!

ASV OTU dada2 qiime2 16S • 429 views
0
Entering edit mode
7 months ago
AfinaM ▴ 30

You can run pre-merged paired end reads as single end reads with qiime dada2 denoise-single. Check out this link: https://docs.qiime2.org/2020.11/plugins/available/dada2/denoise-single/

You can import the files by using qiime tools import --show-importable-formats or qiime tools import --show-importable-types.

Are you planning to run both pre-merged and post-merged reads together using DADA2? You can check out QIIME2 forum which is really helpful if you are interested using DADA2 and/or QIIME2.

0
Entering edit mode

Oh! I saw that you can run single-end reads with q2 dada2, but I wasn't sure if that creates any obfuscated errors. I'm trying to pin down why exactly my PI says you can't generate ASVs from pre-merged paired-end reads...

Would it just be the lack of an explicit option for that use case? Because it doesn't look like there's an obvious prohibition on using pre-merged reads with dada2, but I'm not sure that constitutes an endorsement. (I'm new, though. Maybe you know that's not a concern at all and I'm worrying over nothing! Haha.)

Thank you, by the way, I'll check out the q2 forum, too! :)

0
Entering edit mode

I believe it is not a problem at all if you want to use DADA2 for pre-merged paired-end reads and generates ASV. ASV is obtained from the sample not the reads anyway so all is good. We will just need to consider the parameter used to run DADA2, e.g. sequence length, truncate/trimming.

I have read in q2 forum before about this kind of case and they suggest to assume the sample as single-ends and run with qiime dada2 denoise-single.