Early-life oral microbiome development is a complex community assembly process that influences long-term health outcomes. Nevertheless, microbial functions and interactions driving these ecological processes remain poorly understood. In this study, we analyze oral microbiomes from a longitudinal cohort of 24 mother-infant dyads at 1 and 6 months postpartum using shotgun metagenomics. We identify two previously undescribed Streptococcus and Rothia species to be among the most prevalent, abundant and strongly co-occurring members of the oral microbiome of six-month-old infants. By leveraging metagenome-assembled genomes (MAGs) and genome-scale metabolic models (GEMS) we reveal their genomic and functional characteristics relative to other infant-associated species and predict their metabolic interactions within a network of co-occurring oral taxa. Our findings highlight unique functional features, including genes encoding adhesins and carbohydrate-active enzymes (CAZymes). Metabolic modeling identified potential exchange of key amino acids, particularly ornithine and lysine, between these species, suggesting metabolic cross-feeding interactions that may explain their co-abundance across infant oral microbiomes. Overall, this study provides key insights into the functional adaptations and microbial interactions shaping early colonization in the oral cavity, providing testable hypotheses for future experimental validation.

BACKGROUND: Data from microbiomes from multiple niches is often collected, but methods to analyse these often ignore associations between niches. One interesting case is that of the oral microbiome. Its composition is receiving increasing attention due to reports on its associations with general health. While the oral cavity includes different niches, multi-niche microbiome data analysis is conducted using a single niche at a time and, therefore, ignores other niches that could act as confounding variables. Understanding the interaction between niches would assist interpretation of the results, and help improve our understanding of multi-niche microbiomes. METHODS: In this study, we used a machine learning technique called latent Dirichlet allocation (LDA) on two microbiome datasets consisting of several niches. LDA was used on both individual niches and all niches simultaneously. On individual niches, LDA was used to decompose each niche into bacterial sub-communities unveiling their taxonomic structure. These sub-communities were then used to assess the relationship between microbial niches using the global test. On all niches simultaneously, LDA allowed us to extract meaningful microbial patterns. Sets of co-occurring operational taxonomic units (OTUs) comprising those patterns were then used to predict the original location of each sample. RESULTS: Our approach showed that the per-niche sub-communities displayed a strong association between supragingival plaque and saliva, as well as between the anterior and posterior tongue. In addition, the LDA-derived microbial signatures were able to predict the original sample niche illustrating the meaningfulness of our sub-communities. For the multi-niche oral microbiome dataset we had an overall accuracy of 76%, and per-niche sensitivity of up to 83%. Finally, for a second multi-niche microbiome dataset from the entire body, microbial niches from the oral cavity displayed stronger associations to each other than with those from other parts of the body, such as niches within the vagina and the skin. CONCLUSION: Our LDA-based approach produces sets of co-occurring taxa that can describe niche composition. LDA-derived microbial signatures can also be instrumental in summarizing microbiome data, for both descriptions as well as prediction.

MOTIVATION: The human microbiome plays a key role in health and disease. Thanks to comparative metatranscriptomics, the cellular functions that are deregulated by the microbiome in disease can now be computationally explored. Unlike gene-centric approaches, pathway-based methods provide a systemic view of such functions; however, they typically consider each pathway in isolation and in its entirety. They can therefore overlook the key differences that (i) span multiple pathways, (ii) contain bidirectionally deregulated components, (iii) are confined to a pathway region. To capture these properties, computational methods that reach beyond the scope of predefined pathways are needed. RESULTS: By integrating an existing module discovery algorithm into comparative metatranscriptomic analysis, we developed metaModules, a novel computational framework for automated identification of the key functional differences between health- and disease-associated communities. Using this framework, we recovered significantly deregulated subnetworks that were indeed recognized to be involved in two well-studied, microbiome-mediated oral diseases, such as butanoate production in periodontal disease and metabolism of sugar alcohols in dental caries. More importantly, our results indicate that our method can be used for hypothesis generation based on automated discovery of novel, disease-related functional subnetworks, which would otherwise require extensive and laborious manual assessment. AVAILABILITY AND IMPLEMENTATION: metaModules is available at https://bitbucket.org/alimay/metamodules/ CONTACT: [email protected] or [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: Massively parallel sequencing allows for rapid sequencing of large numbers of sequences in just a single run. Thus, 16S ribosomal RNA (rRNA) amplicon sequencing of complex microbial communities has become possible. The sequenced 16S rRNA fragments (reads) are clustered into operational taxonomic units and taxonomic categories are assigned. Recent reports suggest that data pre-processing should be performed before clustering. We assessed combinations of data pre-processing steps and clustering algorithms on cluster accuracy for oral microbial sequence data. RESULTS: The number of clusters varied up to two orders of magnitude depending on pre-processing. Pre-processing using both denoising and chimera checking resulted in a number of clusters that was closest to the number of species in the mock dataset (25 versus 15). Based on run time, purity and normalized mutual information, we could not identify a single best clustering algorithm. The differences in clustering accuracy among the algorithms after the same pre-processing were minor compared with the differences in accuracy among different pre-processing steps. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online. CONTACT: [email protected] or [email protected]