DADA2: The filter removed all reads for some samples

Hello Sirong,

I think you are doing a good job, and are on the right track. This is good:

Let's calculate the area of overlap you would expect based on these trimming parameters. (Here's how this calculation works, if that's helpful.)

length of amplicon = end (reverse primer) - start (forward primer)
length of amplicon = 806 - 338
length of amplicon = 468

(forward read) + (reverse read) - (length of amplicon) = overlap
277 forward + 203 reverse - 468 amplicon = overlap
-38 = overlap

So, a negative overlap means there is no overlap at all! Instead, you have a gap between reads of 38 basepairs.

This explains why no reads passed the filter.

To get your reads to join, you will have increase your truncation length so the reads overlap a little bit. Try running dada2 again, with new truncation settings, like these:

  --p-trunc-len-f 277 \
  --p-trunc-len-r 255 \

You are correct that it's usually best to trim off low quality regions before joining. But your V3-V4 region is very long, so your reads must also be long if you want to join them.

If the area of overlap is too low quality to join, you can still continue your analysis using just your forward reads! Because the quality of your forward read1 is very good, consider using dada2 denoise-single, which does not require a second read or any joining step.

Let us know what you try next. We are always here to help.

2 Likes