Predicting age from the transcriptome of human dermal fibroblasts

Area of Science:

• Genomics and Bioinformatics
• Aging Research
• Computational Biology

Background:

• Biomarkers of aging are crucial for assessing individual health and understanding age-related diseases.
• Transcriptome profiles, specifically RNA sequencing (RNA-seq), offer a molecular snapshot of cellular activity.
• Existing methods for predicting biological age from transcriptomic data have limitations in accuracy and scope.

Purpose of the Study:

• To investigate if aging signatures are encoded within the human dermal fibroblast transcriptome.
• To develop a robust machine learning model for predicting chronological age from RNA-seq data.
• To assess the model's performance against existing age prediction methods and its utility in studying progeria.

Main Methods:

• Generated genome-wide RNA-seq profiles from 133 human dermal fibroblast samples across a wide age range (1-94 years).
• Developed an ensemble machine learning approach to predict age based on transcriptomic signatures.
• Validated the predictive model on a cohort of ten progeria patients exhibiting accelerated aging.

Main Results:

• The ensemble machine learning model accurately predicted age with a median error of 4 years.
• This performance surpassed that of previously reported age prediction methodologies.
• The model uniquely identified accelerated aging signatures in progeria patients, confirming its sensitivity.

Conclusions:

• Transcriptomic data from human dermal fibroblasts contain robust signatures of biological aging.
• The developed ensemble machine learning method provides a highly accurate tool for age prediction.
• This approach holds promise for studying aging mechanisms and age-related pathologies, including progeria.