Chronological age prediction based on DNA methylation: Massive parallel sequencing and random forest regression

Area of Science:

• Forensic epigenetics
• Genomic analysis

Background:

• DNA methylation (DNAm) analysis is a growing field in forensic science.
• Predicting chronological age from DNAm patterns offers valuable forensic insights.

Purpose of the Study:

• To develop and validate an accurate age-prediction tool for whole blood samples.
• To utilize massive parallel sequencing (MPS) and machine learning for DNAm-based age estimation.

Main Methods:

• Selected 15 age-dependent DNAm markers, refined to 13 based on MPS analysis.
• Trained a random forest algorithm on 208 whole blood samples and validated on 104 samples (ages 18-69).
• Investigated neighboring CpG sites and single nucleotide polymorphisms (SNPs) for improved accuracy.

Main Results:

• The developed tool achieved a mean absolute deviation (MAD) of 3.21 years and RMSE of 3.97 years on the training set.
• Validation on the test set yielded comparable results with MAD of 3.16 years and RMSE of 3.93 years.
• A reduced model with 4 markers also demonstrated reliable age prediction capabilities.

Conclusions:

• A reliable and accurate epigenetic tool for age prediction in whole blood was created.
• The tool is not restricted to linear DNA methylation changes with age.
• Identified novel age-dependent DNA methylation sites for future model enhancements.