while I'm struggling to do beta diversity, I found that there is a problem on my table after denoise with DADA2. I double checked my manifest file and metadata file, I reassure that there are 9samples in total in both of them with exact information and file path. But after denoise-single, 3 of them are gone. Only 6 samples remain in table after dada2 denoise and 3 are gone. Below is my code for denoise.
Can you please provide your sequence quality plot from demux summarize? This will help us understand the nature of the read length and quality, prior to quality control. Thanks!
I would try truncating at around 400 nt. Once done, run feature-table summarize on the resulting table and share here, we can take a look at the results together.
PS - these reads are super long, what sequencing platform are they from?
Your reply helps me a lot! May I ask you one more question related to quality?
I checked the quality at first and I agonized between leaving reads with low quality but long and reads with high quality but short. As you said, if I cut reads at 400nt, the overall quality is improved but half of them are gone. Doesn't it have problem? If it's up to the case, how can I determine it??
Thank you!
P.S the platform noticed on metadata is LS454. How long are mostly reads ?
You could try experimenting with several sets of DADA2 parameters and then compare some downstream results computed with those different outputs, such as diversity and taxonomic classification to see how these parameters are impacting your particular study.
Thank you again, your wonderful reply!! I have one last question of denoising with dada2.
I tried dada2 denoise-single procedure and the results have all samples but most of them have 0 sequences per samples. Is it because standard quality I set for trimming is too high??
Are you using dada2 denoise-single? You have 454 data so you want to be using the dada2 denoise-pyro instead since that is designed with 454 reads in mind.
