PALM

PALM prioritizes genetic risk variants from GWAS summary statistics by integrating cell-type and tissue-specific functional annotations to improve interpretation of non-coding and polygenic association signals.

Key Features:

• Annotation integration: Integrates cell-type/tissue-specific functional annotations with GWAS summary statistics.
• Non-linear modeling: Uses a tree ensemble model to capture non-linear relationships between functional annotations and variant association status.
• Functional gradient-based EM: Fits the tree-based non-linear model using a novel functional gradient–based expectation-maximization algorithm.
• Scalability: Enables efficient fitting across millions of genetic variants and hundreds of functional annotations while maintaining modeling stability.
• Statistical performance: Demonstrated robust control of false discovery rate and enhanced statistical power in simulation studies.
• Annotation prioritization: Produces importance rankings of functional annotations to aid variant prioritization.

Scientific Applications:

• Large-scale GWAS integration: Applied to integrate summary statistics from 30 GWASs with 127 cell type/tissue-specific functional annotations.
• Variant discovery and interpretation: Identified an increased number of prioritized risk variants, including non-coding variants, and ranked annotation contributions to GWAS signals.

Methodology:

Integrates cell-type/tissue-specific functional annotations with GWAS summary statistics using a tree ensemble to model non-linear annotation–association relationships and is fitted via a functional gradient–based expectation-maximization algorithm.