Machine Learning on Human Muscle Transcriptomic Data for Biomarker Discovery and Tissue-Specific Drug Target...

Area of Science:

• Molecular biology
• Gerontology
• Biomarker discovery

Background:

• Significant progress in understanding molecular mechanisms of muscle aging.
• Lack of accessible, tissue-specific biomarkers for evaluating anti-aging interventions.

Purpose of the Study:

• Develop a method for tracking age-related changes in human skeletal muscle.
• Identify tissue-specific biomarkers of aging.
• Validate the predictive accuracy of these biomarkers.

Main Methods:

• Analysis of publicly available gene expression profiles from young and old healthy donors.
• Differential gene expression and pathway analysis.
• Application of supervised machine learning algorithms, including neural networks.

Main Results:

• Confirmed known aging mechanisms: dysregulation of Ca2+ homeostasis, PPAR signaling, neurotransmitter recycling, IGFR, and PI3K-Akt-mTOR signaling.
• Developed a predictive model with 0.91 Pearson correlation and 6.19 years mean absolute error on test data.
• Achieved 0.80 accuracy in age bin prediction on an external validation set (GTEx project).

Conclusions:

• The developed biomarkers accurately predict skeletal muscle age.
• These biomarkers can identify novel molecular targets for anti-aging therapies.
• This approach offers a tool for assessing therapeutic interventions for muscle aging.