I have 16S V3-V4 amplicon sequences using 341F–805R primers. Samples are extracted from soil. The raw data have barcodes and primers, so I use cutadapt to remove it.
Here is the data after barcodes and primers removal:
trimmed-seqs.qzv (321.6 KB)
When I run DADA2, I lose too many reads after filtering and left less than 20% after chimera removal.
qiime dada2 denoise-paired \ --i-demultiplexed-seqs outputs/trimmed-seqs.qza \ --p-trunc-len-f 244 \ --p-trunc-len-r 244 \ --p-n-threads 0 \ --o-table outputs/table.qza \ --o-representative-sequences outputs/rep-seqs.qza \ --o-denoising-stats outputs/denoising-stats.qza
denoising-stats.qzv (1.2 MB)
I can see I have pretty good read quality both the forward and the reverse, so I didn't trunc too much, also to make sure it can overlap. But the biggest problem is I loss too many reads in the filtering step. I searched some forum posts, but their problem like low reads quality or strict
–p-trunc-q value doesn't suit my questions.
I have no idea what causes the problem, any help will be much appreciated!