Sequence counts dropped dramatically after dada2 but looks OK after demultiplexing

[cdc (Yanan Wang)]

Hello,
I am running over 200 samples sequenced by MiSeq, for V3-V4.
The sequence reads from demux.qzv, min 8583, median 70557. However, after dada2, the sequence counts dropped dramatically!!! min 0, median 8, max, 166.
That is really strange. Does anyone know why?

My command for dada2: demux.qzv (307.7 KB)
table-vis.qzv (403.1 KB)
rep-seqs-dada2.qza (113.5 KB)

qiime dada2 denoise-paired --i-demultiplexed-seqs demux-paired-end.qza --p-trunc-len-f 240 --p-trunc-len-r 200 --o-representative-sequences rep-seqs-dada2.qza --p-chimera-method consensus --p-n-threads 0 --o-table table.qza --output-dir dada2

[jwdebelius (Justine Debelius)]

Hi @cdc,

I’d recommend checking out your DADA2 stats, and seeing where you’re losing sequences. Without seeing that my guess is that because of your read length, youre having issues joining your ends. You could try leaving one of your sequences longer so you have more overlap, or you could try DADA2 on forward reads only.

Best,
Justine

[cdc (Yanan Wang)]

Hi Justine,

Thanks for your reply. I have checked the dada2 stats, and it seems like very low number of reads have been merged (hope I interpreted the stats correctly).

What could be the reasons? Why they are not merging? I think my trunc parameters (240 for f and 200 for r) are reasonable based on the QS scores...

stats.tsv (8.8 KB)

[jwdebelius (Justine Debelius)]

Hi @cdc,

I think your parameters are okay based on your quality score, but you may not have enough overlap for a good merge. Your options are to either let yourself have a longer overlap, which may mean losing more sequences during QC, or denoising with forward reads only.

Best,
Justine