Academic literature on the topic '16S rRNA profiling'

Background: Profiling the microbiome of low-biomass samples is challenging for metagenomics since these samples often contain DNA from other sources, such as the host or the environment. The usual approach is sequencing specific hypervariable regions of the 16S rRNA gene, which fails to assign taxonomy to genus and species level. Here, we aim to assess long-amplicon PCR-based approaches for assigning taxonomy at the genus and species level. We use Nanopore sequencing with two different markers: full-length 16S rRNA (~1,500 bp) and the whole rrn operon (16S rRNA–ITS–23S rRNA; 4,500 bp). Methods: We sequenced a clinical isolate of Staphylococcus pseudintermedius, two mock communities (HM-783D, Bei Resources; D6306, ZymoBIOMICS™) and two pools of low-biomass samples (dog skin from either the chin or dorsal back), using the MinION™ sequencer 1D PCR barcoding kit. Sequences were pre-processed, and data were analyzed using the WIMP workflow on EPI2ME or Minimap2 software with rrn database. Results: The full-length 16S rRNA and the rrn operon were used to retrieve the microbiota composition at the genus and species level from the bacterial isolate, mock communities and complex skin samples. For the Staphylococcus pseudintermedius isolate, when using EPI2ME, the amplicons were assigned to the correct bacterial species in ~98% of the cases with the rrn operon marker, and in ~68% of the cases with the 16S rRNA gene. In both skin microbiota samples, we detected many species with an environmental origin. In chin, we found different Pseudomonas species in high abundance, whereas in dorsal skin there were more taxa with lower abundances. Conclusions: Both full-length 16S rRNA and the rrn operon retrieved the microbiota composition of simple and complex microbial communities, even from the low-biomass samples such as dog skin. For an increased resolution at the species level, using the rrn operon would be the best choice.

The aim of this work was to analyze and compare the bacterial communities of 663 samples from a Brazilian hospital by using high-throughput sequencing of the 16S rRNA gene. To increase taxonomic profiling and specificity of 16S-based identification, a strict sequence quality filtering process was applied for the accurate identification of clinically relevant bacterial taxa. Our results indicate that the hospital environment is predominantly inhabited by closely related species. A massive dominance of a few taxa in all taxonomic levels down to the genera was observed, where the ten most abundant genera in each facility represented 64.4% of all observed taxa, with a major predominance of Acinetobacter and Pseudomonas. The presence of several nosocomial pathogens was revealed. Co-occurrence analysis indicated that the present hospital microbial network had low connectedness, forming a clustered topology, but not structured among groups of nodes (i.e., modules). Furthermore, we were able to detect ecologically relevant relationships between specific microbial taxa, in particular, potential competition between pathogens and non-pathogens. Overall, these results provide new insight into different aspects of a hospital microbiome and indicate that 16S rRNA sequencing may serve as a robust one-step tool for microbiological identification and characterization of a wide range of clinically relevant bacterial taxa in hospital settings with a high resolution.

Background: Profiling the microbiome of low-biomass samples is challenging for metagenomics since these samples are prone to contain DNA from other sources (e.g. host or environment). The usual approach is sequencing short regions of the 16S rRNA gene, which fails to assign taxonomy to genus and species level. To achieve an increased taxonomic resolution, we aim to develop long-amplicon PCR-based approaches using Nanopore sequencing. We assessed two different genetic markers: the full-length 16S rRNA (~1,500 bp) and the 16S-ITS-23S region from the rrn operon (4,300 bp). Methods: We sequenced a clinical isolate of Staphylococcus pseudintermedius, two mock communities and two pools of low-biomass samples (dog skin). Nanopore sequencing was performed on MinION™ using the 1D PCR barcoding kit. Sequences were pre-processed, and data were analyzed using EPI2ME or Minimap2 with rrn database. Consensus sequences of the 16S-ITS-23S genetic marker were obtained using canu. Results: The full-length 16S rRNA and the 16S-ITS-23S region of the rrn operon were used to retrieve the microbiota composition of the samples at the genus and species level. For the Staphylococcus pseudintermedius isolate, the amplicons were assigned to the correct bacterial species in ~98% of the cases with the16S-ITS-23S genetic marker, and in ~68%, with the 16S rRNA gene when using EPI2ME. Using mock communities, we found that the full-length 16S rRNA gene represented better the abundances of a microbial community; whereas, 16S-ITS-23S obtained better resolution at the species level. Finally, we characterized low-biomass skin microbiota samples and detected species with an environmental origin. Conclusions: Both full-length 16S rRNA and the 16S-ITS-23S of the rrn operon retrieved the microbiota composition of simple and complex microbial communities, even from the low-biomass samples such as dog skin. For an increased resolution at the species level, targeting the 16S-ITS-23S of the rrn operon would be the best choice.

BackgroundMicrobiome studies of the lower airways based on bacterial 16S rRNA gene sequencing assess microbial community structure but can only infer functional characteristics. Microbial products, such as short-chain fatty acids (SCFAs), in the lower airways have significant impact on the host's immune tone. Thus, functional approaches to the analyses of the microbiome are necessary.MethodsHere we used upper and lower airway samples from a research bronchoscopy smoker cohort. In addition, we validated our results in an experimental mouse model. We extended our microbiota characterisation beyond 16S rRNA gene sequencing with the use of whole-genome shotgun (WGS) and RNA metatranscriptome sequencing. SCFAs were also measured in lower airway samples and correlated with each of the sequencing datasets. In the mouse model, 16S rRNA gene and RNA metatranscriptome sequencing were performed.ResultsFunctional evaluations of the lower airway microbiota using inferred metagenome, WGS and metatranscriptome data were dissimilar. Comparison with measured levels of SCFAs shows that the inferred metagenome from the 16S rRNA gene sequencing data was poorly correlated, while better correlations were noted when SCFA levels were compared with WGS and metatranscriptome data. Modelling lower airway aspiration with oral commensals in a mouse model showed that the metatranscriptome most efficiently captures transient active microbial metabolism, which was overestimated by 16S rRNA gene sequencing.ConclusionsFunctional characterisation of the lower airway microbiota through metatranscriptome data identifies metabolically active organisms capable of producing metabolites with immunomodulatory capacity, such as SCFAs.

Human saliva can be separated by centrifugation into cell pellet and cell-free supernatant, which are called cellular phase and liquid phase in this study. While it is well documented that the cellular phase of saliva contains hundreds of oral bacteria species, little is known whether the liquid phase of saliva contains any information related to oral microbiota. In this study, we analyzed the bacterial nucleic acid contents of the liquid phase of saliva. Using primers universal to most eubacterial 16S rDNA, we detected large amounts of bacterial 16S rRNA and rDNA in the cell-free phase of saliva. Random sequencing analysis of forty PCR amplicons from the cell-free phase of saliva led to 15 operational taxonomic unit (OTU) groups. Furthermore, using denaturing gradient gel electrophoresis (DGGE), we compared 16S rRNA/rDNA profiles derived from liquid phases and cellular phases of saliva samples, and found positive correlations (Pearson Correlation=0.822,P<0.001) between these sample groups. These findings indicate that the liquid phase of saliva contains numerous bacterial 16S rRNA/rDNA molecules that have correlations with bacteria existing in the cellular phase.

The profiling of bacterial communities by the sequencing of housekeeping genes such as that encoding the small subunit ribosomal RNA has revealed the extensive diversity of bacterial life on earth. Standard protocols have been developed and are widely used for this application, but individual habitats may require modification of methods. This review discusses the sequencing and analysis methods most appropriate for the study of the bacterial component of the human oral microbiota. If possible, DNA should be extracted from samples soon after collection. If samples have to be stored for practical reasons, precautions to avoid DNA degradation on freezing should be taken. A critical aspect of profiling oral bacterial communities is the choice of region of the 16S rRNA gene for sequencing. The V1-V2 region provides the best discrimination between species of the genus Streptococcus, the most common genus in the mouth and important in health and disease. The MiSeq platform is most commonly used for sequencing, but long-read technologies are now becoming available that should improve the resolution of analyses. There are a variety of well-established data analysis pipelines available, including mothur and QIIME, which identify sequence reads as phylotypes by comparing them to reference data sets or grouping them into operational taxonomic units. DADA2 has improved sequence error correction capabilities and resolves reads to unique variants. Two curated oral 16S rRNA databases are available: HOMD and CORE. Expert interpretation of community profiles is required, both to detect the presence of contaminating DNA, which is commonly present in the reagents used in analysis, and to differentiate oral and nonoral bacteria and determine the significance of findings. Despite advances in shotgun whole-genome metagenomic methods, oral bacterial community profiling via 16S rRNA sequence analysis remains a valuable technique for the characterization of oral bacterial populations.

ABSTRACTLarge-scale and in-depth characterization of the intestinal microbiota necessitates application of high-throughput 16S rRNA gene-based technologies, such as barcoded pyrosequencing and phylogenetic microarray analysis. In this study, the two techniques were compared and contrasted for analysis of the bacterial composition in three fecal and three small intestinal samples from human individuals. As PCR remains a crucial step in sample preparation for both techniques, different forward primers were used for amplification to assess their impact on microbial profiling results. An average of 7,944 pyrosequences, spanning the V1 and V2 region of 16S rRNA genes, was obtained per sample. Although primer choice in barcoded pyrosequencing did not affect species richness and diversity estimates, detection ofActinobacteriastrongly depended on the selected primer. Microbial profiles obtained by pyrosequencing and phylogenetic microarray analysis (HITChip) correlated strongly for fecal and ileal lumen samples but were less concordant for ileostomy effluent. Quantitative PCR was employed to investigate the deviations in profiling between pyrosequencing and HITChip analysis. Since cloning and sequencing of random 16S rRNA genes from ileostomy effluent confirmed the presence of novel intestinal phylotypes detected by pyrosequencing, especially those belonging to theVeillonellagroup, the divergence between pyrosequencing and the HITChip is likely due to the relatively low number of available 16S rRNA gene sequences of small intestinal origin in the DNA databases that were used for HITChip probe design. Overall, this study demonstrated that equivalent biological conclusions are obtained by high-throughput profiling of microbial communities, independent of technology or primer choice.

The human microbiome - defined as the microbial communities that live in and on our bodies - is emerging as a key factor in human diseases. The expanding research field that investigates the role of the microbiome on human cancer development, termed oncobiome, has led to important discoveries such as the role of Fusobacterium nucleatum in colorectal cancer carcinogenesis and tumor progression. Motivated by these discoveries, this thesis studied the oncobiome from different perspectives, investigating whether alterations to microbial profiles were associated with disease status or an adverse response to treatment. We used both biopsy tissue samples and 16S rRNA amplicon sequencing (N = 36), as well as privately and publicly available fecal whole metagenomes (N = 764) to investigate microbiome-colorectal cancer (CRC) associations. We observed significant increases in species richness in CRC, regardless of sample type or methodology, which was partially due to expansions of species typically from the oral cavity, as well as an overabundance of specific taxa such as Bacteroides fragilis, Fusobacterium, Desulfovibrio and Bilophila in CRC. Functional potential analysis of CRC metagenomes revealed that the choline trimethylamine-lyase (cutC) gene was over-abundant in CRC, with the strength of association dependent on four identified sequence variants, pointing at a novel potential mechanism of CRC carcinogenesis. Predictive microbiome signatures trained on the combination of multiple datasets showed very high and consistent performances on distinct cohorts (average AUC 0.83, minimum 0.81). To investigate the microbiomes role in response to treatment, we profiled microbial communities of gastric wash samples in gastric cancer patients (N = 36) before and after neoadjuvant chemotherapy through 16S rRNA amplicon sequencing. Gastric wash microbial communities presented remarkably high inter-individual variation, with significant decreases in richness and phylogenetic diversity after treatment and associations with pH, pathological response and sample collection. The most abundant genera found in patients before or after chemotherapy treatment included Streptococcus, Prevotella, Rothia and Veillonella. Despite limitations inherent to differing experimental choices, this thesis provides microbiome signatures that can be the basis for clinical prognostic tests and hypothesis-driven mechanistic studies, as well as supporting the role of the human oral microbiome in whole-body diseases.
O microbioma humano - definido como as comunidades microbianas que vivem sobre e dentro do corpo humano - está se tornando um fator cada vez mais importante em doenças humanas. O campo de estudo que investiga o papel do microbioma no desenvolvimento do câncer humano, denominado oncobioma, está crescendo e já levou a importantes descobertas como o papel da espécie Fusobacterium nucleatum na carcinogênese e progressão tumoral de tumores colorretais. Motivado por estas descobertas, esta tese de doutorado analisou o oncobioma por diferentes perspectivas, investigando se alterações nos perfis microbianos estavam associados à presença da doença ou a uma resposta adversa ao tratamento. Usamos tanto amostras de tecidos de biópsias e o sequenciamento do gene 16S rRNA (N = 36), quanto metagenomas fecais públicos e privados (N = 764), para investigar associações entre o microbioma e o câncer colorretal (CCR). Observamos um aumento significativo da riqueza microbiana no CCR, independentemente do tipo da amostra ou metodologia, que era em parte, devido ao aumento de espécies tipicamente presentes na cavidade oral. Observamos também um aumento da abundância de táxons específicos no CCR, que incluíam Bacteroides fragilis, Fusobacterium, Desulfovibrio e Bilophila. Analisando o potencial funcional dos metagenomas, encontramos um aumento significativo da enzima liase colina trimetilamina (cutC) no CCR, cuja associação era dependente de 4 variantes de sequência, demonstrando ser um possível novo mecanismo de carcinogênese no CCR. Assinaturas preditivas do microbioma treinadas na combinação dos estudos demonstraram ser altamente preditivas e consistentes nos diferentes estudos (média de AUC 0.83, mínimo de 0.81). Para investigar o possível papel do microbioma na resposta ao tratamento, analisamos os perfis microbianos do suco gástrico de pacientes com câncer gástrico (N = 36) antes e depois do tratamento quimioterápico neoadjuvante. As comunidades microbianas apresentaram uma variabilidade inter-individual notavelmente grande, com diminuições significativas na riqueza e diversidade filogenética pós tratamento, além de estarem associadas principalmente ao pH, mas também à resposta patológica e ao tempo da coleta. Os gêneros mais abundantes encontrados nos pacientes antes ou depois da quimioterapia incluíam Streptococcus, Prevotella, Rothia e Veillonella. Apesar das limitações inerentes às escolhas experimentais, esta tese proporciona assinaturas do microbioma que podem servir de base para testes clínicos prognósticos e estudos mecanísticos, além de dar mais suporte ao papel do microbioma oral em doenças humanas.

Tese de mestrado Bioinformática e Biologia Computacional, Universidade de Lisboa, Faculdade de Ciências, 2018
As tecnologias de sequenciação paralela massiva têm contribuído imenso para os desenvolvimentos nas áreas que estudam amostras biológicas complexas, como a metataxonómica e a metagenómica. Um dos passos fundamentais na análise de dados derivados de amostras complexas é a determinação da origem taxonómica dos fragmentos genómicos sequenciados. Apesar da diversidade de ferramentas existente, os métodos de identificação taxonómica aplicados produzem resultados variáveis e com um grau de fiabilidade que é difícil de verificar, especialmente a níveis taxonómicos mais baixos, como espécie. Neste estudo, três estratégias de classificação baseadas em correspondência exacta de k-mers, implementadas nos programas Kraken, CLARK e Centrifuge, que são frequentemente utilizadas para classificação de dados de metagenómica, foram aplicadas a dados simulados de metataxonómica, criados a partir de sequências do gene 16S rRNA presentes em bases de dados de referência, de forma a avaliar e comparar a eficiência dessas estratégias de classificação. Esse desempenho foi também comparado com uma estratégia de classificação tipicamente aplicada em processos de metataxonómica e baseada em alinhamento de sequências. Cada estratégia foi avaliada através da classificação de sequências de 14 subregiões do gene 16S rRNA e tendo como referência três bases de dados e versões condensadas dessas bases de dados. No total, as estratégias baseadas em correspondência exacta de k-mers foram avaliadas com um total de 60 combinações de classificador, base de dados e subregião e a estratégia baseada em alinhamento de sequências foi avaliada com 45 combinações das mesmas variáveis. De entre as estratégias baseadas em correspondência exacta de k-mers, a estratégia baseada em correspodência de k-mers discriminativos destacou-se pelo desempenho superior que demonstrou ao classificar as sequências nos níveis taxonómicos de família, género e espécie. A classificação de sequências ao nível de família através de k-mers discriminativos foi, em média, 8.9% (exactidão de 0.878) mais exacta do que a estratégia baseada em alinhamentos de sequências (0.789) e ao nível taxonómico de género ambas as estratégias apresentaram um valor de exactidão igual (0.704). Ao nível de espécie, a classificação baseada em correspondência de k-mers discriminativos apresentou um desempenho superior às restantes estratégias, atingindo um valor médio de exactidão de aproximadamente 0.908 e superior a 0.846 para os 25% de valores de exactidão mais altos. Em termos comparativos, para a segunda melhor estratégia, também baseada em k-mers, os 25% de valores de exactidão mais elevados variaram entre 0.332-0.705. De acordo com os valores absolutos de exactidão, as sub-regiões V4/V3-V4, V4/V3-V4 e V2/V1-V2 foram as que permitiram atingir os valores mais altos de exactidão aos níveis taxonómicos de familia, género e espécie, respectivamente. De forma a explorar novos métodos que possam contribuir para melhorias na área de classificação taxonómica, a estratégia que apresentou melhor desempenho na avaliação das estratégias de classificação, baseada em k-mers discriminativos, foi utilizada para classificar sequências das sub-regiões V2-V4 e V6-V8. Os termos taxonómicos partilhados por sequências de sub-regiões diferentes foram utilizados para modelar as relações taxonómicas em grafos. Dois algoritmos de acoplamento (máximo peso e algoritmo húngaro) e um algoritmo que determina correspondências de muitos para muitos, com base na maioria de termos partilhados, foram utilizados para determinar correspondências entre os nós de cada grafo bipartido, que maximizassem o número de termos taxonómicos comuns entre os nós de cada correspondência determinada. A aplicação de dois valores de cutoff a um subconjunto das correspondências que capturavam a maior parte das diferenças taxonómicas previstas pelo classificador provocou uma diminuição do valor médio de exactidão para a sub-região V2-V4, ao nível de espécie, entre 0.73% e 1.99%, dependendo do algoritmo utilizado para determinar as correspondências. O mesmo procedimento levou a um aumento do valor médio da exactidão de 2.48%, ao nível de espécie, para a sub-região V6-V8, se as correspondências fossem determinadas sempre com base nos casos que partilhavam o maior número de termos taxonómicos. A aplicação dos valores de cutoff a correspondências determinadas pela regra de maioria de termos partilhados permitiu detectar sete espécies que não tinham sido detectadas através das previsões iniciais, diminuindo o número de espécies não detectadas de 31 para 24 (num total de 111 espécies). Uma abordagem alternativa consistiu na reclassificação das sequências representadas pelos nós das correspondências que apresentavam incompatibilidades através do BLAST. A classificação deste subconjunto de sequências permitiu melhorar o valor médio de exactidão, ao nível de espécie para a sub-região V2-V4, entre 1.12% e 1.53%, dependendo do algoritmo a partir do qual as correspondências tinham sido determinadas. Os valores de exactidão para a sub-região V6-V8 aumentaram entre 5.58% e 7.13%. O valor de bitscore calculado para cada previsão do BLAST foi utilizado para alterar os termos taxonómicos associados a cada nó, de forma a favorecer as classificações previstas com um bitscore mais alto. Esta estratégia levou a uma ligeira diminuição no valor de exactidão para ambas as sub-regiões. A reclassificação de sequências com o BLAST reduziu o número de espécies não detectadas, de 31 para um mínimo de 13 ou 16, dependendo se as correspondências eram determinadas através do algoritmo húngaro ou da regra de maioria de termos partilhados, respectivamente. A alteração da informação taxonómica por favorecimento de valores de bitscore mais elevados permitiu detectar mais duas espécies, quando aplicado em correspondências determinadas através do algoritmo húngaro, e mais seis espécies, quando as correspondências tinham sido determinadas pela regra de maioria de termos partilhados. A combinação de sequências representadas por nós acoplados pelo algoritmo húngaro ou pela regra de maioria de termos partilhados levou a grupos multilocus em que as sequências constituintes partilhavam todos os termos taxonómicos com uma exactidão de 0.925 e 0.937, respectivamente. Os resultados apresentados nesta dissertação demonstram que uma estratégia baseada em k-mers e desenvolvida para classificação de dados de metagenómica pode atingir níveis de desempenho considerados superiores em relação às estratégias testadas neste e noutros estudos para a classificação de dados de metataxonómica. Apesar de existirem algumas limitações inerentes à estratégia aqui aplicada, o nível de desempenho observado foi positivo e demonstra que existe potencial nesta abordagem para melhoramentos adicionais. A classificação de sequências de dois loci através de dois classificadores com estratégias diferentes, e a combinação dos resultados de ambos, evidencia as vantagens da combinação das previsões de mais do que um classificador, para minimizar as falhas da aplicação de uma única estratégia de classificação. Para além disso, a combinação dos resultados poderá representar melhorias em relação às estratégias actualmente aplicadas se forem desenvolvidas pontuações estatisticamente mais robustas que indiquem, com elevado nível de confiança, quais as previsões mais correctas, de forma a capturar os casos em que cada sub-região apresenta maior resolução taxonómica e promover a correcta alteração das classificações taxonómicas através de algoritmos de acoplamento. Em conclusão, este trabalho constitui a base para o possível desenvolvimento de uma estratégia que poderá trazer melhoramentos aos métodos de classificação actualmente utilizados.
High-throughput sequencing technologies have greatly contributed to developments in the areas of metataxonomics and metagenomics, that encompass the study of complex microbial samples. One of the fundamental steps in the analysis of data derived from the sequencing of such samples is the taxonomic identification of sequencing reads. Despite great availability and diversity of classification tools, taxonomic classification methods deliver variable results with a degree of reliability that is difficult to determine, especially at lower taxonomic ranks, such as species. In this study, three k-mer based classification strategies commonly used to classify metagenomics sequencing data, implemented in Kraken, CLARK and Centrifuge, were applied to metataxonomics simulated data, generated from 16S rRNA gene sequences of reference databases, in order to assess classification performance against an alignment-based strategy typically applied in metataxonomics pipelines. Each classification strategy was evaluated with three distinct reference databases and their condensed variations, and by classifying sequences of 14 subregions of the 16S rRNA gene. Performance was evaluated over 60 and 45 combinations of classifier, database and subregion for the k-mer based strategies and for the alignment-based strategy, respectively. From the k-mer based strategies, the one based on discriminative k-mers stood out and displayed superior performance at the evaluated ranks of family, genus and species. At family-level, classification based on discriminative k-mers was 8.9% (accuracy of 0.878) more accurate than the alignment-based strategy (0.789) and at genus level both strategies achieved the same mean accuracy value (0.704). At species level, classification based on discriminative k-mers displayed superior performance, achieving a maximum accuracy value of ~0.908 and the top 25% of accuracy values obtained with all combinations of classifier, database and subregion were above 0.846. By contrast, the second best strategy, also k-mer based, reached an accuracy maximum of 0.705 and the top 25% of values started at an accuracy value of 0.332. Regarding absolute accuracy values, greater taxonomic resolution was obtained with the classification of sequences from the V4/V3-V4, V4/V3-V4 and V2/V1-V2 subregions for the family, genus and species levels, respectively. To explore new approaches that might contribute to improvements in classification strategies, the best performing strategy, based on discriminative k-mers, was used to classify sequences from the V2-V4 and V6-V8 subregions. Taxonomic relations between sequences of both loci were modeled through graph structures. Two graph matching algorithms (maximum weight algorithm and hungarian algorithm) and one simple majority rule algorithm were applied to find matches in each bipartite graph that sought to maximize the similarity between each pair of matched nodes. The application of two cutoff values, determined through the distributions of confidence and gamma values from CLARK’s classifications, to a subset of the matches that captured most of discrepancies between taxonomic predictions of both loci caused a drop in the mean accuracy for the V2-V4 subregion, at species level, between 0.73% and 1.99%, depending on the algorithm used to determine matches. For the V6-V8 subregion, an increase in the mean accuracy at species level of 2.48% was achieved for matches determined through the majority rule algorithm. Seven new species, that were not detected in the initial predictions, were detected by altering classification based on the cutoff values, which decreased the number of undetected species from 31 to 24 (in a total of 111 species) when applied to the matches of a simple majority rule algorithm. An alternative approach classified the sequences represented by nodes in the same matches through BLAST. The classification of this subset of sequences improved the mean accuracy at species level for the V2-V4 subregion between 1.12% and 1.53%, depending on the algorithm used to determine matches. Accuracy values for the V6-V8 subregion increased between 5.58% and 7.13%. Using the bitscore of each BLAST prediction to favor high scoring predictions in each match and seek further improvements, led to a slight decrease in the accuracy values for both subregions. Reclassifying a subset of the sequences with BLAST reduced the number of undetected species from 31 to a minimum of 13 and 16, for the hungarian and the majority rule algorithms, respectively. Altering classification based on BLAST’s bitscore allowed the detection of two and six additional species, for the Hungarian and majority rule algorithms, respectively. Combining sequences from both loci according to the matches determined with the hungarian algorithm or the majority rule algorithm led to multilocus groups in which the constituent sequences shared all taxonomic terms with a mean accuracy of 0.925 and 0.937, respectively. The results presented in this dissertation demonstrate that a k-mer based strategy developed for metagenomics data may achieve superior performance in the classification of metataxonomics data and that the potential of that strategy can be further explored. In conclusion, these results show that the combination of classification results from two loci and two distinct classifiers constitutes an approach that can bring a considerable improvement to taxonomic classification. This work hence lays the foundation for the development of a strategy that might result in important gains to currently used methods.

This study aims to measure the impact of protein and amino acid fermentation on the composition and metabolic output of gut microbiota. Although dissimilatory pathways have been described for most amino acids, microbial degradation routes within the gut microbiota are relatively unexplored. The objectives were (1) to characterize amino acid breakdown by the colonic microbiota, (2) to determine the fermentation products formed from individual amino acids/protein (3) to examine how amino acid metabolism is impacted by the presence of a fermentable fiber (prebiotic inulin) and finally (4) to evaluate with an in vivo model (trout fish) diet:- microbe interactions and the development of gut microbiota during fish farming. Interactions between the healthy human intestinal microbiota of the distal colon and different combinations of nutrients were simulated using in vitro pH-controlled anaerobic batch cultures of human faeces. Combining high-throughput sequencing of 16S rRNA amplicons, with high-throughput 1 H NMR, changes in faecal microbiota composition and metabolic output were measured. During exogenous substrate microbial fermentation (e.g. beef, Trp or fish feed) in the large bowel bioactive compounds (harmful or beneficial) are produced. Many factors affect the gut-microbial metabolism including pH, type and quantity of growth substrate (e.g. protein/carbohydrate) and make up of the gut microbiota. Considerable interindividual variation was observed in response to different digested substrates but over all, the beneficial impact of prebiotic fiber fermentation on production of bioactive compounds from amino acids/proteins was confirmed in this study. In trout, although our dietary intervention with essential oils had little impact on the gut microbiota, the study showed for the first time a dramatic shift in the composition and diversity of the gut microbiota in juvenile compared to adult fish. These observations may have relevance in designing dietary strategies to reduce chronic diseases like colon cancer and heart disease and for fish farming respectively