- Open Access
Diversity and enterotype in gut bacterial community of adults in Taiwan
- Chao Liang†1,
- Han-Chi Tseng†2,
- Hui-Mei Chen†1,
- Wei-Chi Wang3,
- Chih-Min Chiu3,
- Jen-Yun Chang3,
- Kuan-Yi Lu3,
- Shun-Long Weng5, 6, 7,
- Tzu-Hao Chang8,
- Chao-Hsiang Chang9,
- Chen-Tsung Weng3,
- Hwei-Ming Wang3 and
- Hsien-Da Huang1, 4, 10Email author
© The Author(s). 2017
- Published: 25 January 2017
Gastrointestinal microbiota, particularly gut microbiota, is associated with human health. The biodiversity of gut microbiota is affected by ethnicities and environmental factors such as dietary habits or medicine intake, and three enterotypes of the human gut microbiome were announced in 2011. These enterotypes are not significantly correlated with gender, age, or body weight but are influenced by long-term dietary habits. However, to date, only two enterotypes (predominantly consisting of Bacteroides and Prevotella) have shown these characteristics in previous research; the third enterotype remains ambiguous. Understanding the enterotypes can improve the knowledge of the relationship between microbiota and human health.
We obtained 181 human fecal samples from adults in Taiwan. Microbiota compositions were analyzed using next-generation sequencing (NGS) technology, which is a culture-independent method of constructing microbial community profiles by sequencing 16S ribosomal DNA (rDNA). In these samples, 17,675,898 sequencing reads were sequenced, and on average, 215 operational taxonomic units (OTUs) were identified for each sample. In this study, the major bacteria in the enterotypes identified from the fecal samples were Bacteroides, Prevotella, and Enterobacteriaceae, and their correlation with dietary habits was confirmed. A microbial interaction network in the gut was observed on the basis of the amount of short-chain fatty acids, pH value of the intestine, and composition of the bacterial community (enterotypes). Finally, a decision tree was derived to provide a predictive model for the three enterotypes. The accuracies of this model in training and independent testing sets were 97.2 and 84.0%, respectively.
We used NGS technology to characterize the microbiota and constructed a predictive model. The most significant finding was that Enterobacteriaceae, the predominant subtype, could be a new subtype of enterotypes in the Asian population.
- 16S rDNA
- Next-generation sequencing
- Gut microbiome
- Predictive model
Microorganisms inhabit various sites of the human body . The largest number of microorganisms is found in the gut . The gut microbiome is associated with human health . For example, the gastrointestinal microbiome affects human physiological functions such as immune function and inflammation suppression, food decomposition and nutrient absorption, regulation of blood substrate via the nervous and/or endocrine system, and recovery rate from bacterial infection . However, some of the underlying mechanisms remain unclear. Enterotypes of the human gut microbiome are not associated with gender, age, or body weight but are influenced by long-term dietary habits. Therefore, we aimed to identify the enterotypes of adults in Taiwan by next-generation sequencing (NGS).
Since the Human Microbiome Project (HMP) was launched by the National Institutes of Health in 2008, NGS has been widely used to study the human microbiome . One of the benefits of NGS is that it is a culture-independent method that can be used to characterize microbial community profiles by sequencing of 16S ribosomal DNA (rDNA). In addition, hundreds to thousands of bacteria can be identified at a time on sequencing 16S rDNA by NGS. Thus, variations in bacteria among different samples can be determined by comparing their quantitative profiles .
In 2011, three enterotypes of the human gut microbiome were identified from 261 human fecal samples from European individuals. The major bacteria in these enterotypes were Bacteroides, Prevotella, and Ruminococcus . This finding was subsequently validated by another approach using the same HMP dataset . However, after identifying the long-term dietary habits in subjects, another study only observed Bacteroides and Prevotella enterotypes in their dataset and reported that Ruminococcus was an ambiguous enterotype . The HMP dataset included gut microbiota that was rich in saturated fats and animal protein, whereas the latter study included microbiota from individuals with plant-based diets that were low in meat and high in carbohydrates . In 2012, two other groups also reported that Ruminococcus could not be clearly classified in their datasets, and Firmicutes were identified as the dominant species in those studies [10, 11].
Based on the data from these previous studies, we were interested in determining if Ruminococcus is an enterotype in the gut microbiota of Taiwanese individuals. To this end, 181 human fecal samples from adults in Taiwan were collected, and the V4 regions of the 16S rDNA gene were sequenced through paired 150-cycle reads using the Illumina MiSeq system. A total of 17,675,898 sequencing reads were sequenced in 181 samples, and 215 operational taxonomic units (OTUs) were identified in each sample on average. The most abundant bacteria identified in the fecal samples were Bacteroides, Prevotella, and Enterobacteriaceae, and their correlation with dietary habits was confirmed. A decision tree model of these three enterotypes was constructed. The accuracies of this model in training and independent testing sets were 97.2 and 84.0%, respectively.
The most significant finding in our study was the identification of Enterobacteriaceae as one of the predominant subtypes in the gut microbiota. This species may be a new subtype of enterotypes in the Asian population.
Sequencing data statistics
We conducted 17,675,898 sequencing reads on 181 stool samples. After filtering the sequences that did not fit the criteria, we further analyzed 16,474,959 sequencing reads. After taxonomy assignment, 9,133,183 sequencing reads were aligned to genes in the 16S rDNA database that had a sequence similarity of at least 97%; 215 OTUs for each sample were identified on average. Detailed information on the sequencing reads is listed in Additional file 1: Table S1.
Enterotype identification in the fecal samples
Summary of optimal cluster numbers
Characteristics of the enterotypes
The questionnaire given to the study subjects included questions about three major determinants regarding the samples. The first was the shape of feces, which was categorized by participants according to the Bristol stool scale. Scores of 1–3 represented “hard” stool, a score of four represented “mid,” and scores of 5–7 represented “watery” that had a high water content . The second determinant was the frequency of excretion. At least one excretion every 2 days was designated “D1+,” excretion two to three times a week was designated “D05,” and excretion once a week or less referred to as “constipation.” The third variable was “protein type,” which referred to the major source of protein in daily diets: the non-red meat group included individuals who eat beans/vegetables, fish, and poultry and the red meat group included individuals who mostly eat livestock.
Association between enterotypes and various other factors from the questionnaire
(n = 30)
(n = 36)
(n = 40)
Gender (global p = 0.047)
Type 1 vs Type 2
Type 1 vs Type 3
Type 2 vs Type 3
Type 2 vs (Type 1 + Type 3)a
Protein (global p = 0.015)
Type 1 vs Type 2
Type 1 vs Type 3
Type 2 vs Type 3
Type 1 vs. (Type 2 + Type 3)a
Shape (global p = 0.014)
Type 1 vs Type 2
Type 1 vs Type 3
Type 2 vs Type 3
(Type 1 + Type 2) vs Type 3a
Stool (global p = 0.064)
Type 1 vs Type 2
Type 1 vs Type 3
Type 2 vs Type 3
Type 1 vs (Type 2 + Type 3)a
Enterotype pathway enrichment analysis
Enterotype 1 shows higher pathway activity than enterotype 2 or enterotype 3 in some KEGG pathways (ko00902, ko00909, ko05168, ko05416, ko05145, ko05210, ko04115, ko04610) (Additional file 3: Table S2). Two metabolic pathways are related to terpenoid biosynthesis; three pathways are related to infections such as virus and parasite; two pathways are associated with cancer and p53 DNA repair system; ko04610 is related to innate immune system.
Classification of enterotypes
Performance of classification model in training sets and independent testing sets
Significant genus lists categorized by enterotype-related metadata
T2 > T1 = T3
T3 > T1 > T2
T1 > T2 = T3
T1 > T2 = T3
T1 > T3
T1 > T3
T1 > T2
T1 > T2 = T3
D1 + < D05
D1 + < D05
Hard > Watery
Hard > Watery
Red > non-red
Red < non-red
Red > non-red
Enterotype-related phenotypes provide data for observing the gastrointestinal tract with the nature of continual flux , e.g., nutrient substrate, water context, or transition status. With regard to stool frequency, the amount of Dialister and Akkermansia in “D05” was higher than that in “D1 + .” According to previous studies, the amount of Dialister was higher in individuals with a high protein diet  and the amount of Akkermansia was higher in those with a fiber-free diet . With regard to shape, the amounts of Parabacteroides and Prevotella in hard stool were higher than those in watery stool. Previous studies also showed that the amount of Prevotella was higher in ethnic groups that had a high fiber diet and lower in ethnic groups that adopted a Western diet . With regard to protein type, the red meat group had abundance of Bifidobacterium and Akkermansia and the non-red meat group had abundance of Megamonas. Higher levels of lipid in the diet increased the amount of Bifidobacterium because it has the ability to digest lipids . Enterotype 1 lacks of predominant bacteria such as Prevotella and Bacteroide, which may lead to a functional imbalance or a potential infectious risk via KEGG pathways (Additional file 3: Table S2).
Microflora, enterotype-related phenotypes, and short-chain fatty acids (SCFAs) were theoretically interwoven as a large association network . Our study validates the connections among those factors (Additional file 5: Figure S2). SCFAs are byproducts of dietary fiber fermentation through microbiota, and they predominantly include acetic, propionic, and butyric acids. SCFAs can promote the growth of bacteria and can be absorbed by humans. Different types of SCFAs are sources of energy in different organs and are associated with intestinal diseases. Several factors control SCFA production in the gut, such as the amount and type of bacteria and the food retention time. The amounts of SCFAs affect pH of the intestine, for example, a higher concentration of SCFAs leads to lower pH. The pH value is associated with the composition of the bacterial community. The complex interaction network in the gut includes the amount of SCFAs, pH value of the intestine, and composition of the bacterial community.
Our results provide a predictive model for further analysis and new insights into enterotypes. An individual may change his/her enterotype by making dietary changes because the characteristics of enterotypes depend on an individual’s dietary habits. Although some researchers pointed that the gut microbiome should not category as ‘Enterotypes or Faecotypes’ since there is no clearly separation among clusters . The classification may be blurred, yet the different features are still there. Thus, knowing one’s enterotype may allow doctors to outline the best diet for patients and to prescribe the most effective drugs.
Feces sample collection
The 181 human feces samples used in this population-based study were collected by Sigma-Transwab (Medical Wire). Feces were temporarily stored at 4 °C before DNA extraction. The exclusion criteria were age less than 10 years, a history of gastrointestinal tract surgery, and hospitalization or antibiotic treatment within the past 2 months. Of the resulting study cohort of 181 individuals, 106 provided complete information on the questionnaire and 75 omitted some information.
In this case study, fresh feces were obtained from participants, and DNA was directly extracted from stool samples using the QIAamp DNA Stool Mini Kit (Qiagen). A swab sample was vigorously vortexed and incubated at room temperature for 1 min. Then, the sample was transferred to a microcentrifuge tube containing 560 μl Buffer ASL, vortexed, and incubated at 37 °C for 30 min. Following this, the suspension was incubated at 95 °C for 15 min, vortexed, and centrifuged at 14,000 rpm for 1 min into pellet stool particles. Extraction was performed following the protocol of the QIAamp DNA Stool Mini Kit. The DNA was eluted with 50 μl Buffer AE and centrifuged at 14,000 rpm for 1 min, after which the DNA extract was stored at −20 °C until further analysis.
Library construction and sequencing of the V4 region of the 16S ribosomal DNA
The PCR primers F515 (5′-GTGCCAGCMGCCGCGGTAA-3′) and R806 (5′-GGACTACHVGGGTWTCTAAT-3′) were designed to amplify the V4 region of the bacterial 16S ribosomal DNA as described previously . PCR amplification was performed in a 50-μl reaction volume containing 25 μl 2X Taq Master Mix (Thermo Scientific), 0.2 μM of forward and reverse primer, and 20 ng DNA template. The reaction process increased the initial temperature to 95 °C for 5 min, followed by 30 cycles of 95 °C for 30 s, 54 °C for 1 min, and 72 °C for 1 min as well as a final extension of 72 °C for 5 min. Next, amplified products were checked by 2% agarose gel electrophoresis and ethidium bromide staining. Amplicons were purified using the AMPure XP beads (Agencourt) and quantified using the Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific), all according to the respective manufacturers’ instructions. For V4 library preparation, Illumina adapters were attached to the amplicons using the Illumina TruSeq DNA Sample Preparation v2 Kit. Purified libraries were processed for cluster generation and sequencing using the MiSeq system.
Filtering 16S rDNA sequencing data for quality
Sequencing reads from different samples were identified and separated according to specific barcodes at the 5’ end of the sequence (two mismatches allowed). The FASTX-Toolkit was employed to process the raw read data files. There were three steps used for sequence quality processing: (i) The command was “fastq_quality_filter –Q33 − q 20 − p 70.” “−q 20” meant that he minimum quality score to be maintained is 20. “−p 70” meant that the minimum percent of bases must have “−q” quality over or equal to 70%. (ii) The command was “fastq_quality_trimmer − t 20 − l 100 − Q33.” “−t 20” meant that bases with lower quality (<20) would be trimmed (checking from the end of the sequence). “−l 100” meant that the minimum acceptable length of sequence was 100 after trimming the sequence. (iii) Sequences were retained if both forward and reverse sequencing reads passed the first and second steps.
Taxonomy assignment for bacteria 16S rDNA sequence
To generate taxonomy assignment, the collection of 16S rDNA sequences was retrieved from the SILVA ribosomal RNA sequence database (release 115) . These sequences were extracted using V4 forward and reverse primers. Then, UCLUST was used to create representative sequence clusters over or equal to 97% similarity . Bowtie2 was used to align sequencing reads against the clusters of the V4 sequence. A 97% similarity standard was applied to the V4 sequence clusters.
Bacterial community analysis
After taxonomy assignment, an OTU table was generated. To normalize the sample size of all samples, a rarefaction process was applied to the OTU table. There are three steps in deciphering the enterotype of stool samples . The first step is to calculate the distance matrix of β-diversity. R package “vegan”  and Python software “Pycogent”  were employed to calculate nine types of matrices, namely Alternative Gower, Bray Cutris, Jaccard, Kulczynski, Chebyshev, Pearson, Horn, Euclidean, and Weighted UniFrac. The second step is to use these matrices as the input data for three cluster algorithms: HC, k-means clustering, and PAM methods. For PAM, there were two types of inputs: one was the distance matrix of β-diversity and the other was the point information of XY axes that were transformed from the distance matrix. R package was also used to perform the clustering process. The third step is to evaluate the quality of clustering results. Silhouette score was calculated by R package “clusterSim” . A higher score represents better quality of clustering results. To explore the association between bacterial community and factors related to individuals, which were extracted from the questionnaires, weighted α-diversity (Shannon index), chi-square test, and analysis of variance (ANOVA) were performed with R package. There were three criteria for identifying significant bacteria in the groups: the first was relative abundance > 1% in at least one group, the second was fold change in relative abundance between two groups ≥ log2(3) or ≤ log2(1/3). The third was p value ≤ 0.05. In order to construct a predictive model for classifying the three enterotypes of the stool samples, 181 stool samples were separated into two sets: the training set contained 106 samples, which were from individuals who provided complete information on the questionnaire, and the independent testing set contained 75 samples, which were from those who did not provide complete information. The decision tree, which was a rule-based machine learning method, was used to construct the predictive model for the three enterotypes. C4.5, which is a well-built decision tree package, was employed to perform this modeling process . Tax4Fun was adopted to the pathway enrichment analysis with ANOVA .
We would like to thank the Come True Biomedical Inc. (CTB) for sharing their pearls of wisdom with us during the course of this research.
This article has been published as part of BMC Genomics Volume 18 Supplement 1, 2016: Proceedings of the 27th International Conference on Genome Informatics: genomics. The full contents of the supplement are available online at http://bmcgenomics.biomedcentral.com/articles/supplements/volume-18-supplement-1.
This project was funded by the Ministry of Science and Technology of the Republic of China [MOST 103-2628-B-009-001-MY3, MOST 104-2627-M-009-008-, MOST 104-2319-B-400-002, MOST 104-2911-I-009-509, MOST 105-2633-B-009-003, MOST 104-2314-B-195-014] and by the Ministry of Health and Welfare of the Republic of China [MOHW105-TDU-B-212-134002]. We also thank Veterans General Hospitals and University System of Taiwan (VGHUST) Joint Research Program [VGHUST 105-G1-4-2]. This paper is particularly supported by “Aiming for the Top University Program” of the National Chiao Tung University and Ministry of Education, Taiwan, R.O.C. The source of funding for the publication of the study is MOST 103-2628-B-009-001-MY3.
Availability of data and materials
HDH, CL, and HCT designed the study. HCT, SLW, CTW, KYL, and CHC lead the clinical study and sample collection. WCW, CMC, JYC, and HMW processed the samples and generated the sequence data. HDH, CL, WCW, and HMC analyzed and interpreted the data. HDH, WCW, THC, and HMC wrote the manuscript. All authors read and approved the final manuscript.
The authors declare that they have no competing interests.
Consent for publication
Ethics approval and consent to participate
Participants were recruited at Tseng Han-Chi General Hospital (Nantou, Taiwan). They volunteered to participate in the study. Before interviewing, the participants were informed or reminded of the purpose of the study and that they may withdraw at any time. Informed consent was obtained from all participants prior to enrolment in the study.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Aziz Q, Dore J, Emmanuel A, Guarner F, Quigley EM. Gut microbiota and gastrointestinal health: current concepts and future directions. Neurogastroenterol Motil. 2013;25(1):4–15.View ArticlePubMedGoogle Scholar
- Franzosa EA, Morgan XC, Segata N, Waldron L, Reyes J, Earl AM, Giannoukos G, Boylan MR, Ciulla D, Gevers D, et al. Relating the metatranscriptome and metagenome of the human gut. Proc Natl Acad Sci U S A. 2014;111(22):E2329–38.View ArticlePubMedPubMed CentralGoogle Scholar
- Korecka A, Arulampalam V. The gut microbiome: scourge, sentinel or spectator? J Oral Microbiol. 2012;4:9367.Google Scholar
- Curtis H, et al. Structure, function and diversity of the healthy human microbiome. Nature. 2012;486(7402):207–14.Google Scholar
- Koren O, Knights D, Gonzalez A, Waldron L, Segata N, Knight R, Huttenhower C, Ley RE. A guide to enterotypes across the human body: meta-analysis of microbial community structures in human microbiome datasets. PLoS Comput Biol. 2013;9(1):e1002863.View ArticlePubMedPubMed CentralGoogle Scholar
- Arumugam M, Raes J, Pelletier E, Le Paslier D, Yamada T, Mende DR, Fernandes GR, Tap J, Bruls T, Batto JM, et al. Enterotypes of the human gut microbiome. Nature. 2011;473(7346):174–80.View ArticlePubMedPubMed CentralGoogle Scholar
- Zhou Y, Mihindukulasuriya KA, Gao H, La Rosa PS, Wylie KM, Martin JC, Kota K, Shannon WD, Mitreva M, Sodergren E, et al. Exploration of bacterial community classes in major human habitats. Genome Biol. 2014;15(5):R66.View ArticlePubMedPubMed CentralGoogle Scholar
- Wu GD, Chen J, Hoffmann C, Bittinger K, Chen YY, Keilbaugh SA, Bewtra M, Knights D, Walters WA, Knight R, et al. Linking long-term dietary patterns with gut microbial enterotypes. Science. 2011;334(6052):105–8.View ArticlePubMedPubMed CentralGoogle Scholar
- De Filippo C, Cavalieri D, Di Paola M, Ramazzotti M, Poullet JB, Massart S, Collini S, Pieraccini G, Lionetti P. Impact of diet in shaping gut microbiota revealed by a comparative study in children from Europe and rural Africa. Proc Natl Acad Sci U S A. 2010;107(33):14691–6.View ArticlePubMedPubMed CentralGoogle Scholar
- Zupancic ML, Cantarel BL, Liu Z, Drabek EF, Ryan KA, Cirimotich S, Jones C, Knight R, Walters WA, Knights D, et al. Analysis of the gut microbiota in the old order Amish and its relation to the metabolic syndrome. PLoS ONE. 2012;7(8):e43052.View ArticlePubMedPubMed CentralGoogle Scholar
- Huse SM, Ye Y, Zhou Y, Fodor AA. A core human microbiome as viewed through 16S rRNA sequence clusters. PLoS ONE. 2012;7(6):e34242.View ArticlePubMedPubMed CentralGoogle Scholar
- Mondot S, de Wouters T, Dore J, Lepage P. The human gut microbiome and its dysfunctions. Dig Dis. 2013;31(3–4):278–85.View ArticlePubMedGoogle Scholar
- Lewis SJ, Heaton KW. Stool form scale as a useful guide to intestinal transit time. Scand J Gastroenterol. 1997;32(9):920–4.View ArticlePubMedGoogle Scholar
- Chassard C, Dapoigny M, Scott KP, Crouzet L, Del’homme C, Marquet P, Martin JC, Pickering G, Ardid D, Eschalier A, et al. Functional dysbiosis within the gut microbiota of patients with constipated-irritable bowel syndrome. Aliment Pharmacol Ther. 2012;35(7):828–38.View ArticlePubMedGoogle Scholar
- Hooda S, Vester Boler BM, Kerr KR, Dowd SE, Swanson KS. The gut microbiome of kittens is affected by dietary protein:carbohydrate ratio and associated with blood metabolite and hormone concentrations. Br J Nutr. 2013;109(9):1637–46.View ArticlePubMedGoogle Scholar
- Jakobsdottir G, Xu J, Molin G, Ahrne S, Nyman M. High-fat diet reduces the formation of butyrate, but increases succinate, inflammation, liver fat and cholesterol in rats, while dietary fibre counteracts these effects. PLoS ONE. 2013;8(11):e80476.View ArticlePubMedPubMed CentralGoogle Scholar
- Pop M, Walker AW, Paulson J, Lindsay B, Antonio M, Hossain MA, Oundo J, Tamboura B, Mai V, Astrovskaya I, et al. Diarrhea in young children from low-income countries leads to large-scale alterations in intestinal microbiota composition. Genome Biol. 2014;15(6):R76.View ArticlePubMedPubMed CentralGoogle Scholar
- Bordoni A, Amaretti A, Leonardi A, Boschetti E, Danesi F, Matteuzzi D, Roncaglia L, Raimondi S, Rossi M. Cholesterol-lowering probiotics: in vitro selection and in vivo testing of bifidobacteria. Appl Microbiol Biotechnol. 2013;97(18):8273–81.View ArticlePubMedGoogle Scholar
- Hijova E, Chmelarova A. Short chain fatty acids and colonic health. Bratisl Lek Listy. 2007;108(8):354–8.PubMedGoogle Scholar
- Jeffery IB, Claesson MJ, O’Toole PW, Shanahan F. Categorization of the gut microbiota: enterotypes or gradients? Nat Rev Microbiol. 2012;10(9):591–2.View ArticlePubMedGoogle Scholar
- Weng SL, Chiu CM, Lin FM, Huang WC, Liang C, Yang T, Yang TL, Liu CY, Wu WY, Chang YA, et al. Bacterial communities in semen from men of infertile couples: metagenomic sequencing reveals relationships of seminal microbiota to semen quality. PLoS ONE. 2014;9(10):e110152.View ArticlePubMedPubMed CentralGoogle Scholar
- Pruesse E, Quast C, Knittel K, Fuchs BM, Ludwig W, Peplies J, Glockner FO. SILVA: a comprehensive online resource for quality checked and aligned ribosomal RNA sequence data compatible with ARB. Nucleic Acids Res. 2007;35(21):7188–96.View ArticlePubMedPubMed CentralGoogle Scholar
- Edgar RC. Search and clustering orders of magnitude faster than BLAST. Bioinformatics. 2010;26(19):2460–1.View ArticlePubMedGoogle Scholar
- Wang J, Linnenbrink M, Kunzel S, Fernandes R, Nadeau MJ, Rosenstiel P, Baines JF. Dietary history contributes to enterotype-like clustering and functional metagenomic content in the intestinal microbiome of wild mice. Proc Natl Acad Sci U S A. 2014;111(26):E2703–10.View ArticlePubMedPubMed CentralGoogle Scholar
- Dixon P. VEGAN, a package of R functions for community ecology. J Veg Sci. 2003;14:927–30.View ArticleGoogle Scholar
- Knight R, Maxwell P, Birmingham A, Carnes J, Caporaso JG, Easton BC, Eaton M, Hamady M, Lindsay H, Liu Z, et al. PyCogent: a toolkit for making sense from sequence. Genome Biol. 2007;8(8):R171.View ArticlePubMedPubMed CentralGoogle Scholar
- Walesiak M. Cluster analysis with ClusterSim computer program and R environment. Acta Universitatis Lodziensis: Folia Oeconomica. 2008;216:303–11.Google Scholar
- Quinlan JR. C4.5: programs for machine learning. San Mateo: Morgan Kaufmann Publishers; 1993.Google Scholar
- Asshauer KP, Wemheuer B, Daniel R, Meinicke P. Tax4Fun: predicting functional profiles from metagenomic 16S rRNA data. Bioinformatics. 2015;31(17):2882–4.View ArticlePubMedPubMed CentralGoogle Scholar