Characterization of microbiota signatures in Iberian pig strains using machine learning algorithms

Area of Science:

• Livestock genetics and microbiome research.
• Animal science and agricultural biotechnology.

Background:

• Host genetics are a major driver of livestock microbiome composition.
• Iberian pigs exhibit genetic and phenotypic variability, impacting meat quality.
• Gut microbiota differences across Iberian pig strains are not well-understood.

Purpose of the Study:

• To explore gut microbiota variations in two Iberian pig strains (Entrepelado and Retinto) and their crosses.
• To utilize machine learning (ML) to identify microbial taxa distinguishing genetic backgrounds.
• To assess the potential application of these findings in the pig industry.

Main Methods:

• Analysis of 16S rRNA gene sequencing data from 237 Iberian pigs.
• Application of nine machine learning algorithms (e.g., Catboost, SVM) to classify genetic groups.
• Beta diversity analysis to assess microbial compositional divergence.

Main Results:

• Machine learning models successfully classified pig genetic groups based on gut microbiota.
• The Support Vector Machine (SVM) model achieved the highest accuracy (AUROC 0.83) for differentiating purebred strains.
• Key discriminating genera identified include Acetitomaculum, Butyricicoccus, and Limosilactobacillus.

Conclusions:

• Gut microbiota composition varies significantly among Iberian pig strains and crosses.
• Machine learning effectively identifies genetic signatures within the pig gut microbiome.
• Identified microbial taxa are linked to lipid metabolism, suggesting a role in fat-related trait variations.