Data processing and Bioinformatics

We processed reads from our VESPA data sets with both QIIME 2¹⁰³ and DADA2 v.1.16.0¹⁰⁴ in the R environment v.3.6.3 and found that, while results were similar, DADA2 was more user-friendly (i.e., did not require installation of new software, required less steps, and was implementable within a familiar computing environment). Read files were converted to vectors and filtered for quality using the filterAndTrim command with default settings plus modifiers to remove primers (trimLeft = c(18,20)), residual PhiX reads (rm.phix = TRUE), and short sequences (minLen = 100). Error rate for forward and reverse reads were calculated using the learnErrors command, data were dereplicated using the derepFastq command, and Sequence Variants were inferred using the dada command. Read pairs were merged using the mergePairs command with justConcatenate = TRUE and chimeras were removed using the removeBimeraDenovo command with default parameters. Taxonomy assignments were made using the assignTaxonomy command and the PR² reference sequence database version 5.1.0 (10.5281/zenodo.7805244), which contains 18S and 16S sequences at species-level resolution. For comparison we also tested two other taxonomy databases (DOI 10.5281/zenodo.1447329): v132 which includes all eukaryotic organisms from the SILVA v132 database and v128 which includes all eukaryotic organisms from the SILVA v128 database plus corrected species labels for Blastocystis and additional Entamoeba sequences. However, we found that the PR² database returned higher numbers of fully assigned ASVs. Any ASVs not assigned taxonomy using the PR² database were queried against the full NCBI nucleotide database on September 3^rd, 2022 using MegaBLAST¹⁰⁵ with default parameters.