Lengths ASVs V3-V4 region

[andrebolerbarros]

Hi everyone,

I am currently working with NovaSeq data, using the modified loess for monotonicity, and trimmed according to what is discussed in the tutorial (trimleft for primers, kept maximum sequence length for trunLen).

I ran DADA2, removed chimeras, kept only ASV's classified as "Bacteria" and also removed unclassified ASVs at the Phylum level. The final distribution of ASVs length is the following:

352 353 410 419 420 421 422 423 424 426 427 429 437 438 439 440 441 442 444 445 446 
  1   1   1   8 163  74  66  89  12   4   3   1  13  19  91 159   6   1   9  73 3

According to this previous issue (#896 (comment)) and references such as this: https://pmc.ncbi.nlm.nih.gov/articles/PMC5785224/, I can safely conclude variation is expected, in a binomial fashion.

Doing some "crude" math, we get:

161nt (V3) + 291nt (V4) = 452nt
452nt - (17 primer fwd + 20 primer rev) = 415nt

186nt (V3) + 291nt (V4) = 477nt
477nt - (17 primer fwd + 20 primer rev) = 440nt

So, the values I get are around ~5nt of difference from the "theoretical" values. If my math is correct, then my results match with the expectation, correct? I am still puzzled with the number of intermediate values...