Dada2 Stats: low sample frequency, non-chimeric reads, and merged metrics

Background: 300 cycle kit, 616F-1132R 18S primers, paired-end demultiplexed reads

Here is my code:
qiime dada2 denoise-paired
--i-demultiplexed-seqs demux-paired-end.qza
--p-trunc-len-f 0
--p-trunc-len-r 0
--o-table dada2_table.qza
--o-representative-sequences dada2_rep_set.qza
--o-denoising-stats dada2_stats.qza

Here is the output for my dada2_stats.qzv post-denoising.
I have very low merged, percentage of input merged, non-chimeric, and percentage of input non-chimeric
values. For comparison, here are the dada2_stats.qzv from Qiime2 Tutorial.

An important point to note: My reads are ~600bp 18S V3V4 reads but a 300 cycle kit was used. could the lack of coverage cause the issue here (no overlap?)

Hi @mkweber,
Unfortunately, the dada2-stats file you linked is not a valid like. Could you send upload the file here or use a dropbox sharable link to share the file?

Lack of overlap could definally be the issue.
Quick math: 1132-616 = 516
So roughly your region is 516bp long, but variation in lengths might cause your region to be shorter or longer than the quick math I did there. 300 bp on the forward and reverse should be good enough.

Here's a screenshot sample from the qzv file:

Here you can see many samples have no reads passing post denoising:

1 Like

Yikes! that is not ideal! Lets keep investigating :mag:

Given the quick math I did above, I am really surprised that we are seeing such low merging success. The only think I can think is there are adapters in your sequence that are preventing the sequences from merging.

Do you know if you have adapters in your sequences?

I'm unsure if I have adaptors -- would a quick check by using the "head" command on a R1 fastq file and grepping the adaptor sequences?

1 Like

If you know that your adapter sequence is, that is a great way to check!

1 Like

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.