Benjamin Callahan

This error is happening in code being called by DADA2, but that is implemented outside the DADA2 package. There is some issue with being able to open one of the...

`assignTaxonomy` makes assignments on a per-sequence basis, it will not be affected by the order of the ASVs in the sequence table at all.

> Is that outcome to be expected? If not, I'd be curious to know why this might be happening. Importantly, there is an element of randomness in the naive Bayesian...

Hm... let me investigate a bit more, but I suspect what is happening is that the initialization state of the C-side random number generators is affected by the sequences previously...

I'm gonna try to at least take a look into this again before our next release (which will be ~Oct). Perfect reproducibility (assuming the random number generator is pre-seeded identically)...

@d4straub As you have correctly surmised, the randomness of the bootstrapping in `assignTaxonomy` is still not controlled by `set.seed` R-side. Very briefly, when we implemented multi-threaded `assignTaxonomy` we had to...

Reading the MAASLIN2 documentation, it seems like you can give it the taxonomic feature table as a data.frame (but that it expects the taxa to be the columns, so transposed...

The workaround right now is to enforce monotonicity in the fitted error model. The error model (`getErrors(errF)`) is just a matrix with columns corresponding to Q=0...41 (usually), and so is...

@jgrembi What we think is happening is that for binned quality scores, there are very few observations at the intermediate consensus quality scores. This isn't shown on these plots, but...

> we do have some mock Zymo NovaSeq, but it is V3-V4; I assume you preferably want V4? Not at all. We are completely ambivalent about the locus sequenced. That...