Denoising of reads and generation of ESV table in amplicon sequencing

[shrutarshi]

I used the below mentioned steps to quality filter and de-replicate reads generated post amplicon sequencing of particular gene

1. Sequencing success and read quality verified using FastQC v0.11.8
2. Merging forward and reverse reads of each sample
3. Forward and reverse primer trimming as well as retaining at least 108bp length reads from each sample
4. Error filtering (error rate 0.5, filter length 100-300bp, min size length-8) in Usearch
5. Finding unique sequences
6. Unoise3 function in USearch (denoising) with alpha value of 5 for generation of ZOTU

My questions are:

1. Do we have to denoise each sample individually or merge all the quality filtered reads from different samples and then denoise it?
2. What are the exact tools and functions required for each process (individual denoising/ multiple merged read denoising)
3. How do we get sample specific abundance for each ESV/ASV/ZOTU, in case we denoise individually?
4. I have nearly 100 samples. How can I parallelly run the same steps on different samples?

Can someone please share a standard workflow for further analyzing (after step “f”) the quality filtered reads till the taxonomy assignment step?