Bioinformatic strategies to address limitations of 16rRNA short-read amplicons from different sequencing platforms

Abstract

Sequencing the 16S gene rRNA has become a popular method when identifying bacterial communities. However, recent studies address differences in the characterization based on sample preparation, sequencing platforms, and data analysis. In this work, we tested some of the available user-friendly protocols for data analysis with the reads obtained from the sequencing machines Illumina MiSeq and Ion Torrent Personal Genome Machine (PGM). We sought for the advantages and disadvantages of both platforms in terms of accuracy, detected species, and abundance, analyzing a staggered mock community. Four different pipelines were applied: QIIME 1.9.1 with default parameters, QIIME 1.9.1 with modified parameters and chimera removal, VSEARCH 2.3.4, and QIIME 2 v.2018.2. To address the limitations of species level detection we used species-classifier SPINGO. The optimal pipeline for PGM platform, was the use of QIIME 1.9.1 with default parameters (QIIME1), except when a study requires the detection of Bacteroides or other Bacteroidaceae members, in which QIIME1MOD (with chimera removal) seems to be a good alternative. For Illumina Miseq, VSEARCH strategy can be a good option. Our results also confirm that all the tested pipelines can be used for metagenomic analysis at family and genus level