Comparative Transcriptome Analysis of Bovine, Porcine, and Sheep Muscle Using Interpretable Machine Learning Models

Area of Science:

• Genomics and Bioinformatics
• Animal Science
• Molecular Biology

Background:

• Muscle growth is crucial for meat quality and economic value in livestock.
• Understanding muscle development genes is vital for improving livestock breeding and meat production.
• Existing research on cross-species muscle development genes requires further systematic investigation.

Purpose of the Study:

• To identify key genes regulating muscle growth and development across different livestock species.
• To perform a cross-species transcriptome analysis using machine learning for gene discovery.
• To uncover homologous and species-specific genes impacting muscle development in cattle, pigs, and sheep.

Main Methods:

• Utilized a dataset of 275 transcriptomes from porcine, bovine, and ovine muscle tissues.
• Employed nine machine learning models, including Support Vector Classifier (SVC) and Support Vector Machine (SVM).
• Applied SHapley Additive exPlanations (SHAP) to analyze transcriptome data and identify significant genes (SHAP genes).

Main Results:

• Identified 41 homologous genes (e.g., NANOG, TLR9) consistently expressed across cattle, pigs, and sheep.
• Discovered species-specific candidate genes for cattle (e.g., SLC47A1), pigs (e.g., DRP2), and sheep (e.g., COL10A1).
• KEGG pathway analysis revealed involvement of ether lipid metabolism, cortisol synthesis, and calcium signaling in muscle development.

Conclusions:

• The study successfully identified key genetic regulators of muscle growth across major livestock species.
• Homologous and species-specific genes provide targets for improving muscle development and meat quality.
• Identified pathways offer insights into the molecular mechanisms underlying muscle growth and development in agricultural animals.