More sequences were discarded from the V4F-V6R than the V6F-V6R dataset, indicating that the sequencing quality of the V4F-V6R dataset was inferior to that of the V6F-V6R. This difference in sequencing quality affected the α-diversity estimations, which will be discussed below. Secondly, we screened the chimeras with UCHIME. Because the sequencing of 101 bp

from both ends could not sequence through the whole V4 to V6 region of the 16S rRNA, we linked each pair of tags with 30 Ns to allow screening of the chimeras. After this step, we acquired 263,127 tags from the V4F-V6R primer set (an average of 9,398 tags per sample) and 714,938 tags from the V6F-V6R primer set (an average of 25,533 tags per sample). Once again, many more chimeras were found with the V4F-V6R Selumetinib chemical structure than the V6F-V6R dataset. This result is reasonable, as the V4 to V6 region (approximately 550 bp) is much longer than the V6 region (approximately 65 bp)

and spans conservative sequences find more in the 16S rRNA, thus being more likely to form chimeras during the process of PCR amplification [17]. Finally, to unify the region and length of the tag, the same 60 bp sequence next to the V6R primer was extracted from both primer sets. To avoid the influence of different sequencing depths, we rarefied all samples to 5,000 tags for a consistent sequencing depth. The Good’s coverage of all samples with 5,000 tags was higher than 0.95 with 0.96 ± 0.005 (mean ± SEM) for samples from the V4F-V6R datasets and 0.98 ± 0.004 for the V6F-V6R datasets, indicating that the sequencing depth was sufficient for reliable analysis of these fecal microbial community samples. Based on these data, analyses including α-diversity (within-community diversity), β-diversity (between-communities diversity), microbial structure and biomarker determination were evaluated,

as they are fundamental for microbiome research. In addition to the quality filtering results, four external standards were sequenced simultaneously with each of the two libraries for a direct comparison of the sequencing quality. The external standards were samples with only one known cloned sequence Bumetanide as the PCR template, and the accuracy was checked at each base position. By LY411575 in vivo comparing the sequencing results of the external standards with the known sequence, we could, to some extent, evaluate the sequencing quality of the library. All external standards were also filtered to remove ambiguous bases (N) and chimeras as above. As shown in Additional file 1: Figure S1, the proportion of sequences which have 100% identity with the external standard in the V6F-V6R library was higher than that of the V4F-V6R library (0.939 vs. 0.879, t-test, P < 0.001), while the proportion of error sequences was significantly lower in the V6F-V6R than the V4F-V6R library, indicating that the sequencing quality of the former was superior to that of the latter.