Maysa-N/LDA-for-Vaginal-Microbiome: Latent Dirichlet Allocation (LDA): A Natural Language Processing Model Applied to the Vaginal Microbiome

The vaginal microbiome is like a giant puzzle, but not every piece fits neatly together. With a Latent Dirichlet Allocation (LDA) model, we uncovered seven hidden sub-communities that help explain why some sows stay healthy while others face discharge and stillbirth. This dataset contains the results of a Latent Dirichlet Allocation (LDA) mixed-membership model applied to the vaginal microbiome of first-parity sows. The model identified seven distinct microbial sub-communities associated with reproductive outcomes, including vaginal discharge and stillbirth occurrence. Sub-community 1 (Lactobacillaceae, Limosilactobacillus) was linked to healthy sows with no discharge or stillbirths, while sub-communities 6 (Enterococcus, Aerococcus, Prevotella) and 7 (Porphyromonas, Ezakiella, Anaerococcus) were associated with severe vaginal discharge. The LDA model provides a framework for understanding heterogeneous microbial community structures and their relationship to reproductive performance in swine