Linking community composition and ecosystem function via the cultivation-independent analysis of marker genes, e.g., the 16S rRNA gene, is a staple of microbial ecology and dependent disciplines. The certainty of results, independent of the bioinformatic handling, is imperative for any advances made within the field. In this work, thermophilic anaerobic co-digestion experimental data, together with primary and waste-activated sludge prokaryotic community data, were analyzed with two pipelines that apply different principles when dealing with technical, sequencing, and PCR biases. One pipeline (VSEARCH) employs clustering methods, generating individual operational taxonomic units (OTUs), while the other (DADA2) is based on sequencing error correction algorithms and generates exact amplicon sequence variants (ASVs). The outcomes of both pipelines were compared within the framework of ecological-driven data analysis. Both pipelines provided comparable results that would generally allow for the same interpretations. Yet, the two approaches also delivered community compositions that differed between 6.75% and 10.81% between pipelines. Inconsistencies were also observed linked to biologically driven variability in the samples, which affected the two pipelines differently. These pipeline-dependent differences in taxonomic assignment could lead to different conclusions and interfere with any downstream analysis made for such mis- or not-identified species, e.g., network analysis or predictions of their respective ecosystem service.
Keywords: ASV; Illumina; OTU; bioinformatics; microbial community composition; microbiome; pipelines; sequence data handling; sequence processing; wastewater treatment.