pyaging

pyaging integrates multiple aging clocks to estimate biological age from molecular measurements including DNA methylation profiles, transcriptomics, histone-mark ChIP-seq, and ATAC-seq for comparative and biomarker analyses across species.

Key Features:

• Integration of diverse aging models: Supports linear models, principal component analysis (PCA) models, neural networks, and automatic relevance determination models to enable selection and comparison of aging clock architectures.
• Comprehensive molecular data support: Accommodates DNA methylation profiles, transcriptomics, histone-mark ChIP-seq, and ATAC-seq for cross-assay and cross-study analyses.
• GPU optimization: Uses a PyTorch-based backend to enable GPU acceleration for inference on large datasets and computationally intensive models.
• Multi-species analysis capability: Supports analyses across multiple species, including mammals and Caenorhabditis elegans.

Scientific Applications:

• Biomarker development: Derives and applies molecular biomarkers that estimate biological age and can stratify age-associated risk.
• Comparative analysis: Benchmarks and compares aging clocks across species and data types to investigate shared and divergent aging signatures.
• Machine learning enablement: Provides a standardized environment for applying and evaluating machine learning models on aging datasets to support model development and reproducible inference.

Methodology:

Provides a unified execution layer for multiple aging clock models and data modalities and employs GPU acceleration via a PyTorch backend; supported model families include linear models, PCA models, neural networks, and automatic relevance determination models.