Limits...
Characterization of bacteria in biopsies of colon and stools by high throughput sequencing of the V2 region of bacterial 16S rRNA gene in human.

Momozawa Y, Deffontaine V, Louis E, Medrano JF - PLoS ONE (2011)

Bottom Line: It was shown (1) that sequence coverage lower than 1,000 reads influenced quantitative and qualitative differences between samples measured by UniFrac distances.In particular, Firmicutes Bacilli were not extracted well by one method. (3) Quantitative and qualitative difference in bacteria from ileum to rectum colon were not observed, but there was a significant positive trend between distances within colon and quantitative differences.Results of human colonic bacteria analyzed using high-throughput sequencing were highly dependent on the experimental design, especially the number of sequence reads, DNA extraction method, and sample type.

View Article: PubMed Central - PubMed

Affiliation: Unit of Animal Genomics, GIGA-Research and Faculty of Veterinary Medicine, University of Liège, Liège, Belgium. Yukihide.Momozawa@guest.ulg.ac.be

ABSTRACT

Background: The characterization of the human intestinal microflora and their interactions with the host have been identified as key components in the study of intestinal disorders such as inflammatory bowel diseases. High-throughput sequencing has enabled culture-independent studies to deeply analyze bacteria in the gut. It is possible with this technology to systematically analyze links between microbes and the genetic constitution of the host, such as DNA polymorphisms and methylation, and gene expression.

Methods and findings: In this study the V2 region of the bacterial 16S ribosomal RNA (rRNA) gene using 454 pyrosequencing from seven anatomic regions of human colon and two types of stool specimens were analyzed. The study examined the number of reads needed to ascertain differences between samples, the effect of DNA extraction procedures and PCR reproducibility, and differences between biopsies and stools in order to design a large scale systematic analysis of gut microbes. It was shown (1) that sequence coverage lower than 1,000 reads influenced quantitative and qualitative differences between samples measured by UniFrac distances. Distances between samples became stable after 1,000 reads. (2) Difference of extracted bacteria was observed between the two DNA extraction methods. In particular, Firmicutes Bacilli were not extracted well by one method. (3) Quantitative and qualitative difference in bacteria from ileum to rectum colon were not observed, but there was a significant positive trend between distances within colon and quantitative differences. Between sample type, biopsies or stools, quantitative and qualitative differences were observed.

Conclusions: Results of human colonic bacteria analyzed using high-throughput sequencing were highly dependent on the experimental design, especially the number of sequence reads, DNA extraction method, and sample type.

Show MeSH

Related in: MedlinePlus

Association between coverage and the number of OTUs.(A) The number of OTUs sampled as a function of number of reads. The data points represent mean ± SD of five randomized samplings. (B) Coverage to detect OTU with different frequencies with ≥95% of confidence. The data points were estimated based on the binomial distribution (see Materials and Methods). The x axis is shown in logarithmic scale.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC3037395&req=5

pone-0016952-g002: Association between coverage and the number of OTUs.(A) The number of OTUs sampled as a function of number of reads. The data points represent mean ± SD of five randomized samplings. (B) Coverage to detect OTU with different frequencies with ≥95% of confidence. The data points were estimated based on the binomial distribution (see Materials and Methods). The x axis is shown in logarithmic scale.

Mentions: The number of OTUs for all reads were 240 and 346 for sample B and C, respectively. After denoise process by AmpliconNoise, the latest version of Pyronoise [10], this number became 237 and 350, respectively. The number of OTUs in two samples, B and C, increased according to coverage and did not reach a plateau, even at 14,000 reads (Figure 2A). Figure 2B shows the coverage necessary to detect a given OTU with ≥95% confidence based on the binominal distribution. In order to detect an OTU present with 0.1% of frequency, >4,000 reads are required. For an OTU with a frequency of 0.01%, >44,000 reads are required.


Characterization of bacteria in biopsies of colon and stools by high throughput sequencing of the V2 region of bacterial 16S rRNA gene in human.

Momozawa Y, Deffontaine V, Louis E, Medrano JF - PLoS ONE (2011)

Association between coverage and the number of OTUs.(A) The number of OTUs sampled as a function of number of reads. The data points represent mean ± SD of five randomized samplings. (B) Coverage to detect OTU with different frequencies with ≥95% of confidence. The data points were estimated based on the binomial distribution (see Materials and Methods). The x axis is shown in logarithmic scale.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC3037395&req=5

pone-0016952-g002: Association between coverage and the number of OTUs.(A) The number of OTUs sampled as a function of number of reads. The data points represent mean ± SD of five randomized samplings. (B) Coverage to detect OTU with different frequencies with ≥95% of confidence. The data points were estimated based on the binomial distribution (see Materials and Methods). The x axis is shown in logarithmic scale.
Mentions: The number of OTUs for all reads were 240 and 346 for sample B and C, respectively. After denoise process by AmpliconNoise, the latest version of Pyronoise [10], this number became 237 and 350, respectively. The number of OTUs in two samples, B and C, increased according to coverage and did not reach a plateau, even at 14,000 reads (Figure 2A). Figure 2B shows the coverage necessary to detect a given OTU with ≥95% confidence based on the binominal distribution. In order to detect an OTU present with 0.1% of frequency, >4,000 reads are required. For an OTU with a frequency of 0.01%, >44,000 reads are required.

Bottom Line: It was shown (1) that sequence coverage lower than 1,000 reads influenced quantitative and qualitative differences between samples measured by UniFrac distances.In particular, Firmicutes Bacilli were not extracted well by one method. (3) Quantitative and qualitative difference in bacteria from ileum to rectum colon were not observed, but there was a significant positive trend between distances within colon and quantitative differences.Results of human colonic bacteria analyzed using high-throughput sequencing were highly dependent on the experimental design, especially the number of sequence reads, DNA extraction method, and sample type.

View Article: PubMed Central - PubMed

Affiliation: Unit of Animal Genomics, GIGA-Research and Faculty of Veterinary Medicine, University of Liège, Liège, Belgium. Yukihide.Momozawa@guest.ulg.ac.be

ABSTRACT

Background: The characterization of the human intestinal microflora and their interactions with the host have been identified as key components in the study of intestinal disorders such as inflammatory bowel diseases. High-throughput sequencing has enabled culture-independent studies to deeply analyze bacteria in the gut. It is possible with this technology to systematically analyze links between microbes and the genetic constitution of the host, such as DNA polymorphisms and methylation, and gene expression.

Methods and findings: In this study the V2 region of the bacterial 16S ribosomal RNA (rRNA) gene using 454 pyrosequencing from seven anatomic regions of human colon and two types of stool specimens were analyzed. The study examined the number of reads needed to ascertain differences between samples, the effect of DNA extraction procedures and PCR reproducibility, and differences between biopsies and stools in order to design a large scale systematic analysis of gut microbes. It was shown (1) that sequence coverage lower than 1,000 reads influenced quantitative and qualitative differences between samples measured by UniFrac distances. Distances between samples became stable after 1,000 reads. (2) Difference of extracted bacteria was observed between the two DNA extraction methods. In particular, Firmicutes Bacilli were not extracted well by one method. (3) Quantitative and qualitative difference in bacteria from ileum to rectum colon were not observed, but there was a significant positive trend between distances within colon and quantitative differences. Between sample type, biopsies or stools, quantitative and qualitative differences were observed.

Conclusions: Results of human colonic bacteria analyzed using high-throughput sequencing were highly dependent on the experimental design, especially the number of sequence reads, DNA extraction method, and sample type.

Show MeSH
Related in: MedlinePlus