16S V4V5 benchmarking study: A comparison of Element and Illumina sequencing platforms at the UMGC

December 16, 2024

The University of Minnesota Genomics Center (UMGC) is evaluating the Element Biosciences AVITI sequencer for its 16S sequencing services. To assess comparability with Illumina sequencers, the UMGC performed 16S V4V5 amplicon sequencing on both the Illumina NextSeq 2000 and the Element AVITI. The primary objective was determining whether transitioning from the NextSeq platform to the AVITI would produce equivalent results.

Materials and Methods:
The study consisted of 1,143 root, rhizosphere, and soil DNA samples. Two sets of serial dilutions (1:8 and 1:64) were made for each sample, and QC was run for both the V4V5 and ITS1 markers. Species and genus count tables were created by uploading the taxa-bar-plots.qzv file for each sequencing project to the QIIME2 viewer (https://view.qiime2.org/) and downloading a .csv file for level 6 or level 7 tables for genera and species, respectively.

Results:
The percent of raw reads passing DADA2 pipeline is similar for AVITI and NextSeq sequencing. Mean and median Phred scores tended to be higher for AVITI sequencing forward reads compared to the NextSeq sequencing, whereas the quality of reverse reads declined more rapidly for AVITI, requiring trimming at 160 bp to allow for comparable numbers of quality reads to be retained by dada2.

In general, more reads were retained by cutadapt for AVITI compared to NextSeq 2000, and more AVITI reads passed dada2 quality filtering. The percent of reads retained by dada2 for each additional step (denoising, merging, and chimera detection) were comparable between AVITI and NextSeq 2000 samples, though slightly more Illumina reads were retained after denoising. The final percentage of raw reads remaining at the end of the dada2 pipeline was comparable between the AVITI and NextSeq 2000.

Detected genera abundances correlated in AVITI and NextSeq sequencing. For individual samples, correlation coefficients varied, with non-diverse root communities showing the smallest correlation coefficients (minimum R = 0.118) and more diverse communities showing significant positive correlations (maximum R = 0.553). At the species and genus level, correlations were greater, with a range of R values from 0.294 to 0.793 for species and 0.613 to 0.883 for genera.

**Figure 2.** Samples with the smallest (left) and largest (right) Pearson correlation statistic (R). Species relative abundances were log10 transformed after adding a pseudocount equal to half of the value. Pearson correlation tests were performed between these species relative abundances detected by the NextSeq 2000 vs. AVITI. The red dashed line is the identity line.

genera_relative_abundances — **Figure 3.** Samples with the smallest (left) and largest (right) Pearson correlation statistic (R). Genera relative abundances were log10 transformed after adding a pseudocount equal to half of the value. Pearson correlation tests were performed between these genus relative abundances detected by the NextSeq 2000 vs. AVITI. The red dashed line is the identity line.

Conclusions:
Optimizing the truncation length parameters for AVITI and NextSeq 2000 sequencing data resulted in comparable percentages of reads being retained and successfully processed by the dada2 pipeline. Correlation testing suggests that the similarity between results obtained by AVITI and NextSeq 2000 sequencing is greatest for genera and species but also comparable for ASVs, especially for more diverse communities. While it is not advisable to switch sequencing methods mid-project, conducting research using the AVITI sequencer is expected to produce results that are comparable to NextSeq 2000 sequencing for microbiome amplicon sequencing projects.

References: