Comparison of library preparation methods
Next, using a 13-sample dataset, we compared sequencing results generated using a prevailing library preparation technique (Illumina Nextera XT) with those created using a new commercially available method which allows for high-throughput preparation at a lower per-sample-price (iGenomx Riptide). Estimates of Shannon diversity between the Nextera XT and iGenomx Riptide generated datasets (both rarefied to 1.6 million bacterial read pairs per sample) were highly correlated (R2 = 0.98, t = 22.988, p< 0.01). Shannon diversity estimates tended to be lower among the sequenced iGenomx Riptide libraries, but this effect was non-significant (t = -1.953 p = 0.08). Bray-Curtis estimates of beta-diversity were also correlated between Nextera XT and iGenomx Riptide created datasets (mantel: r = 0.86, p< 0.01; Figure 3A), as were log-transformed species-level estimates of taxa relative abundances (R2 = 0.97, t = 1121.42, p < 0.01; Figure 3B).