The oral microbiota, the second-largest microbial community in the human body, plays critical roles in systemic health and has been implicated in type 2 diabetes (T2D). To explore the potential causal pathways linking oral microbes, circulating metabolites, and T2D, we conducted a two-sample Mendelian randomization (MR) analysis using East Asian genome-wide association study (GWAS) datasets. Oral microbiome summary statistics were obtained from 2,948 individuals in CNGBdb, plasma metabolite data (136 metabolites) from a Singapore Chinese cohort, and T2D data from a large East Asian meta-analysis (25,079 cases; 29,611 controls). SNPs associated with oral microbial taxa ((p−5) were used as instrumental variables, and causal estimates were derived primarily using inverse-variance weighting, with MR-Egger, weighted median, and MR-PRESSO as sensitivity analyses. Two-step MR identified causal effects of Gemella, Campylobacter_A, and TM7 taxa on T2D through modulation of specific acylcarnitines (e.g., C14:2, C16:1, C18:1), which increased T2D risk. Functional enrichment implicated insulin, PI3K-Akt, AMPK, TGF-$\beta$, and PPAR signaling pathways, suggesting metabolic and immune regulatory mechanisms. Genetic correlation analysis found methionine significantly correlated with T2D$(\text{rg}=0.848, \mathrm{p}=0.036)$, while colocalization provided suggestive lipid-related shared loci, particularly HexCer(d18:1/24:0) and SM(d18:1/16:0). Overall, our results indicate that specific oral microbes may promote T2D development by altering sphingolipid-related metabolism, highlighting potential mechanistic links between oral dysbiosis and metabolic dysfunction. These findings position the oral microbiome as a promising biomarker and therapeutic target in metabolic disease, underscoring the need for validation in larger and multi-ethnic cohorts.

Periodontal disease is common among older adults, with around 50% of all adults aged 30 years and older having some form of gum disease. Although the disease might seem minor, without early or proper diagnosis, the disease can affect all supporting structures of the teeth, including gums, periodontal ligament, and alveolar bone. This research can help find patterns common among people with periodontal disease and possibly prevent the effects of it in the elderly. Here, we explored the deep learning capabilities of convolutional neural networks (CNN) and VGG16 models for detecting disease-associated patterns derived from 16S rRNA gene sequences of plaque samples. The CNN model achieved a 90.2% accuracy for differentiating periodontal disease from health samples, with 76% precision and 86% recall. This model could lead to earlier detection and more effective treatment of periodontal disease, ultimately improving patients’ dental health outcomes and quality of life.

Depression, a prevalent yet often overlooked mood disorder, poses significant challenges in pediatric mental health due to subtle and variable symptom presentations across developmental stages. This study explores the potential of machine learning models—CatBoost and Random Forest—to detect depression in children using tabular data from the National Health Interview Survey (NHIS) (2019–2023). By leveraging consistent survey questions, particularly the DEPFREQ_C variable indicating the frequency of sadness or depression, the models classified children reporting “Daily” as positive cases. The dataset was partitioned into 70-20-10 for training, validation, and testing sets. CatBoost and Random Forest were trained using distinct hyperparameter optimization methods, yielding maximum accuracies of 93.24% and 91.84%, respectively. Performance metrics, including precision, recall, and F1-score, highlight CatBoost’s ability to handle categorical data and survey-specific variables. Additionally, analyses of pre- and post-COVID-19 data revealed minimal impact on model performance. While these models should only complement clinical diagnostics rather than replace them, these findings highlight the potential of machine learning in early-stage detection of depression among children.

Frequency Chaos Game Representation (FCGR) is a method where DNA sequences are converted into a 2D image representing k-mer nucleotide frequencies with additional geometric fractal patterns. We used FCGR to analyze nextgeneration sequences (NGS) of 16S rRNA gene amplicons derived from oral samples with or without caries in combination with machine deep learning to differentiate disease from health. The NGS data captured 16S rRNA genes of all species in the oral cavity for studying oral microbiome, which has been shown to associate with the progression of caries. Our hypothesis is that FCGR patterns of caries and caries-free samples are unique and can be differentiated by deep learning using convolutional neural network model (CNN) - a model specialized in classifying images. Samples labeled as caries and caries-free were split into 80-10-10 for training, testing, and validation purposes in CNN training. An accuracy of 87.5% was achieved in this pilot study. Our study showed that FCGR transforms DNA sequences into fractal image patterns and can be applied to oral microbiome NGS data for classifying clinical oral samples as caries or caries-free. This approach allows for fast and early diagnosis and prevention of major cavities.

MOTIVATION: RNA expression signals detected by high-density genomic tiling microarrays contain comprehensive transcriptomic information of the target organism. Current methods for determining the RNA transcription units are still computation intense and lack the discriminative power. This article describes an efficient and accurate methodology to reveal complicated transcriptional architecture, including small regulatory RNAs, in microbial transcriptome profiles. RESULTS: Normalized microarray data were first subject to support vector regression to estimate the profile tendency by reducing noise interruption. A hybrid supervised machine learning algorithm, hidden Markov support vector machines, was then used to classify the underlying state of each probe to 'expression' or 'silence' with the assumption that the consecutive state sequence was a heterogeneous Markov chain. For model construction, we introduced a profile geometry learning method to construct the feature vectors, which considered both intensity profiles and changes of intensities over the probe spacing. Also, a robust strategy was used to dynamically evaluate and select the training set based only on prior computer gene annotation. The algorithm performed better than other methods in accuracy on simulated data, especially for small expressed regions with lower (<1) SNR (signal-to-noise ratio), hence more sensitive for detecting small RNAs. AVAILABILITY AND IMPLEMENTATION: Detail implementation steps of the algorithm and the complete result of the transcriptome analysis for a microbial genome Porphyromonas gingivalis W83 can be viewed at http://bioinformatics.forsyth.org/mtd.

BACKGROUND: Current commercial high-density oligonucleotide microarrays can hold millions of probe spots on a single microscopic glass slide and are ideal for studying the transcriptome of microbial genomes using a tiling probe design. This paper describes a comprehensive computational pipeline implemented specifically for designing tiling probe sets to study microbial transcriptome profiles. RESULTS: The pipeline identifies every possible probe sequence from both forward and reverse-complement strands of all DNA sequences in the target genome including circular or linear chromosomes and plasmids. Final probe sequence lengths are adjusted based on the maximal oligonucleotide synthesis cycles and best isothermality allowed. Optimal probes are then selected in two stages - sequential and gap-filling. In the sequential stage, probes are selected from sequence windows tiled alongside the genome. In the gap-filling stage, additional probes are selected from the largest gaps between adjacent probes that have already been selected, until a predefined number of probes is reached. Selection of the highest quality probe within each window and gap is based on five criteria: sequence uniqueness, probe self-annealing, melting temperature, oligonucleotide length, and probe position. CONCLUSIONS: The probe selection pipeline evaluates global and local probe sequence properties and selects a set of probes dynamically and evenly distributed along the target genome. Unique to other similar methods, an exact number of non-redundant probes can be designed to utilize all the available probe spots on any chosen microarray platform. The pipeline can be applied to microbial genomes when designing high-density tiling arrays for comparative genomics, ChIP chip, gene expression and comprehensive transcriptome studies.