Comparison to Mothur + Consension
We randomly selected five out of the 61 fungi datasets from Wurzbacher
et al., (2019), ranging from 201 to 447 reads per dataset (Supplementary
Table 1). These cover five fungi species of the genus Inocybe for
ribosomal DNA (rDNA) and the full ribosomal tandem repeat region (TR).
We provide alignments of the corresponding Sanger sequences with our
consensus sequences in the Supplementary (Supplementary files 2-6). In
their approach, Wurzbacher et al., (2019) first perform operational
taxonomic unit (OTU) clustering on the read data using Mothur (Schloss
et al., 2009). Next, they create consensus sequences using Consension
(Wurzbacher et al., 2019).
In general, we see that for both ONT and PacBio data NGSpeciesID and the
Mothur + Consension pipeline perform equally well, generating consensus
sequences with 98.6% to 100% accuracy (Table 1). In three out of the
five cases, the two pipelines produced consensus sequences with the same
accuracy, while in one case each software slightly outperformed the
other (Table 1). Medaka polishing outperformed Racon polishing in four
out of five cases (Table 1).