Meta analysis with different 16S region and get too low features after DADA2

[Jonathan (Hyun)]

Hi @jwdebelius ! Thank you so much for your super detailed and kind comments. They made me think all process again and more in depth.

@jwdebelius

Are you discarding untrimmed reads when you do primer trimming?

Nop, I didn't state the option, "--p-discard-untrimmed" on my command, so I didn't discard untrimmed reads.

@jwdebelius

Have you checked how many reads you're losing at this step?

I haven't checked this.. So you recommend comparing raw-demux.qzv file and primer-filtered-demux.qzv files and check how many reads are filtered out, right? And the smaller/none of reads are filtered out would be the best?

@jwdebelius

I would recommend this for a primer pair, but not for all the studies overall.

Sorry, I don't understand this. So you recommend using paired-end as paired-end (not using only forward) and single-end as single-end? this post said I should use forward read only when I deal with single-end and paired-end studies.

One more question on here: When I denoise them using dada2, I found out samples from one study lose most of their reads on merge steps and I couldn't make them be merged with changing trim/trunc length many times. Would it be fine to use just forward read on this study and use paired-end with other studies?

@jwdebelius

You should not do quality control before you run DADA2. DADA2 assumes that your reads not quality filtered, so I would recommend either using Deblur if you want to quality filter, or not doing that step.

This is my mistake. I'll try again referring to this comment.

@jwdebelius

I would recommend the same parameters within the same hypervariable region, but would suggest that you can do variable regions specific primers.

Is it fine to trim/trunc different parameter if studies are different hypervariable region data?

@jwdebelius

Why are you doing taxonomic assignment after you did closed refernece OTU clustering?

Maybe what I understand on this process is wrong. After OTU clustering, what I can get is OTU files (file which cluster reads analyzed as being from same species somehow; e.g OTU_1, OTU_2, OTU_3...), table.qza which is containing which feature/species(shown like "GQ448970.1.1284" on the table.qzv) exist in how many samples, and rep-seqs.qza which is containing representative sequence(shown like "AB506202.1.1522" on the rep-seqs.qzv) of specific feature. We don't know what "GQ448970.1.1284" is, we don't know whose "AB506202.1.1522" is.
This is why classifier needs and it assign taxonomy on top of those code-like names so that we can finally see through bar-plots with friendly(?) names like E.coli, S.epidermidis. Am I on wrong direction..?

@jwdebelius

Keep asking questions.

This is why I love QIIME2 and this forum. I'm not alone