Error rates could not be estimated -- unsure why

[colinbrislawn (Colin J Brislawn)]

@alexkrohn

I assume I'd run into these same problems even if I used deblur instead of dada2, right? Because deblur also uses the distribution of sequencing errors based on the quality scores?

Not necessarily! Deblur uses a different error model, so it might work just fine with the binned quality scores.

Definitely worth a try!

EDIT: Deblur uses a pretrained error model, which won't mind these binned scores, but also was not built for them. So, idk. More discussion about binned quality scores and deblur in this thread.