Recovery of microbial community profile information hidden in chimeric sequence reads

The next frontier in the field of microbiome studies is identification of all microbes present in the microbiome and accurate determination of their abundance such that microbiome profiles can serve as reliable assessments of health or disease status. PCR-based 16S rRNA gene sequencing and metagenome shotgun sequencing technologies are the prevailing approaches used in microbiome analyses. Each poses a number of technical challenges associated with PCR amplification, sample availability, and cost of processing and analysis. In general, results from these two approaches rarely agree completely with each other. Here, we compare these methods utilizing a set of vaginal swab and lavage specimens from a cohort of 42 pregnant women collected for a pilot study exploring the effect of the vaginal microbiome on preterm birth. We generated the microbial community profiles from the sequencing reads of the V3V4 and V4V5 regions of the 16S rRNA gene in the vaginal swab and lavage samples. For a subset of the vaginal samples from 12 subjects, we also performed metagenomic shotgun sequencing analysis and compared the results obtained from the PCR-based sequencing methods. Our findings suggest that sample composition and complexity, particularly at the species level, are major factors that must be considered when analyzing and interpreting microbiome data. Our approach to sequence analysis includes consideration of chimeric reads, by using our chimera-counting BlastBin program, and enables recovery of microbial content information generated during PCR-based sequencing methods, such that the microbial profiles more closely resemble those obtained from metagenomic read-based approaches.