International interlaboratory study comparing single organism 16S rRNA gene sequencing data: Beyond consensus sequence comparisons

This study presents the results from an interlaboratory sequencingstudy for which we developed a novel high-resolution method for comparing data from different sequencingplatforms for a multi- copy, paralogous gene. The combination of PCR amplification and 16S ribosomal RNAgene (16S rRNA) sequencinghas revolutionized bacteriology by enabling rapid identification, frequently without the need for culture. To assess variability between laboratories in sequencing16S rRNA, six laboratories sequencedthe gene encoding the 16S rRNA from Escherichia coli O157:H7 strain EDL933 and Listeria monocytogenes serovar 4b strain NCTC11994. Participants performed sequencingmethods and protocols available in their laboratories: Sanger sequencing, Roche 454 pyrosequencing®, or Ion Torrent PGM®. The sequencingdata were evaluated on three levels: (1) identity of biologically conserved position, (2) ratio of 16S rRNA gene copiesfeaturing identified variants, and (3) the collection of variant combinations in a set of 16S rRNA gene copies. The same set of biologically conserved positions was identified for each sequencingmethod. Analytical methods using Bayesian and maximum likelihood statistics were developed to estimate variant copyratios, which describe the ratio of nucleotides at each identified biologically variable position, as well as the likely set of variant combinations present in 16S rRNA gene copies. Our results indicate that estimated variant copyratios at biologically variable positions were only reproducible for high throughput sequencingmethods. Furthermore, the likely variant combination set was only reproducible with increased sequencingdepth and longer read lengths. We also demonstrate novel methods for evaluating variable positions when comparing multi- copygene sequencedata from multiple laboratories generated using multiple sequencingtechnologies.