EpiSmokEr2: a robust epigenetic classifier for smoking status inference using Illumina EPIC methylation data

Area of Science:

• Epigenetics
• Biomarkers
• Computational Biology

Background:

• Tobacco smoking causes lasting DNA methylation (DNAm) changes in blood.
• These DNAm changes can function as long-term biomarkers for smoking exposure.

Purpose of the Study:

• To develop and validate a DNA methylation classifier for determining smoking status.
• To create a reliable tool for assessing long-term smoking exposure.

Main Methods:

• Developed EpiSmokEr2, a DNAm classifier using 511 CpGs and LASSO regression on Illumina EPIC array data.
• Trained the model on 1343 samples from the Young Finns Study and validated it across six independent datasets.
• Utilized both EPIC and EPICv2 array platforms for validation.

Main Results:

• EpiSmokEr2 achieved 87% sensitivity and 86% specificity in differentiating current from never smokers.
• Predicted smoking status strongly correlated with established DNAm smoking scores and GrimAge.
• The classifier demonstrated robustness with up to 10% missing CpG data.

Conclusions:

• EpiSmokEr2 is a dependable DNA methylation-based estimator of smoking status.
• The open-source R package facilitates widespread application in epidemiological and clinical studies.
• This tool aids in understanding the biological impact of smoking.