Two many ASV in my dataset

[YuZhang]

Dear all,
After processing the data with DADA2, I found that the number of sequences in the sample was 3W-40W

But the sequencing company said it was normal.
Then I filter the dataset by bacteria

qiime taxa filter-table \
--i-table ../table-dada2-3.qza \
--i-taxonomy ../taxonomy-3.qza \
--p-include d__Bacteria \
--o-filtered-table Bacteria-dada2-table.qza \

# filter sequence
qiime taxa filter-seqs \
  --i-sequences ../rep-seq-dada2-3.qza \
  --i-taxonomy ../taxonomy-3.qza \
  --p-include d__Bacteria \
  --o-filtered-sequences Bacteria-sequences.qza \

qiime feature-table summarize \
  --i-table Bacteria-dada2-table.qza \
  --o-visualization Bacteria-dada2-table.qzv \
  --m-sample-metadata-file ../metadata_qiime2.txt

Bacteria-dada2-table.qzv
Results:

Align to tree

qiime phylogeny align-to-tree-mafft-fasttree \
--i-sequences Bacteria-sequences.qza \
--p-n-threads auto \
--o-alignment aligned-Bacteria-rep-seqs.qza \
--o-masked-alignment masked-aligned-Bacteria-rep-seqs.qza \
--o-tree Bacteria-unrooted-tree.qza \
--o-rooted-tree Bacteria-rooted-tree.qza

According to the search, this error may be due to too many ASVs.

So, what do I need to do with my data after DADA2?
Rarefaction or filter out the lower frequence before rarefaction to reduce the ASV number?

[timanix (Timur Yergaliyev)]

Filtering out features with low overall counts sounds like a good idea to reduce memory requirements for the job.
So, you can filter features with counts less than a certain threshold (50,100 or other numbers), filter sequences based on filtered feature table and then align sequences. In my experience it will decrease memory requirements and speed up the process.

[YuZhang]

Do I need rarefaction after filtering？

[timanix (Timur Yergaliyev)]

Not necessarily. For core-metrics, one need to rarefy the data, but it is already implemented in the plugin. You don't need to rarefy for other plugins unless it is specified in the plugin options.

[YuZhang]

How do I determine the filter criteria?

According this figure？My goal is to study bacteria. Do I need to filter the data into bacteria and then filter out the low quality?

[timanix (Timur Yergaliyev)]

I would do it in that order.
Usually for bacteria I filter out all features with absolute counts less than 50, but you also can filter based on relative abundances (like less than 1%, 0.1%) and prevalence.