The bias generated by sequencing platforms in analyzing of maternal-neonate gut microbiota profiles and diversity

Abstract

Background: Microbiome is an important internal ecosystem closely related to host health. Most of the bacteria existed in the internal ecosystem cannot be isolated with laboratory bacteriological culture methods, while 16S rDNA sequencing is considered and used for the bacterial identification by through the high-throughput platforms. The aim of this study was to compare the microbiota analysis result using two next-generation sequencing platforms and bioinformatics pipelines. Results: 56 maternal-neonate fecal samples were sequenced and analyzed by 16S rRNA amplicon sequencing both by Ion Torrent S5-xl and Illumina Hiseq 2500 with standard protocols at same lab. For the richness and diversity of microbiota, index of chao1, observed_specise, PD_whole_tree, simpson and good_converage varied significantly except Shannon index at two platforms (P<0.05). The relative abundance of bacteria at different taxonomy levels is checked from phylum to species level, the more species of bacteria sequenced and annotated, the lower the correlation of the relative abundance of the bacteria founded between two platforms. The sequencing results are consistent between two platforms. Principal component analysis (PCA) results showed that more than 87% of samples were concentrated. According to principal coordinate analysis (PCoA), 56 samples of the two platforms were divided into two clusters, and the compliance rate of the two platforms is 71.43%. The differences between microbial community structures generated from two platforms were tested by multi-response permutation procedure (MRPP), which showed significant differences at family and genus levels separately (A=0.094, P=0.001; A=0.085, P=0.002). When maternal and neonate samples were considered, at family level, there was no difference in microbiota composition between two platform for maternal group (A=0.006, P=0.149), while in the neonate group, it showed significant differences (A=0.035, P=0.006). At the genus level, there existed significant differences in microbiota both in maternal and neonate group (A=0.0216, P=0.004; A=0.098, P=0.001). Conclusion: Although the relative abundance of microbiota sequenced at two different platforms is basically similar, the diversity and correlation coefficient are still quite different. To increase reproducibility and reliability in cohort studies, it is important to use the same sequencing platforms and the corresponding pipeline to reduce the systematic error in microbiome analysis.