Haplotype function score improves biological interpretation and cross-ancestry polygenic prediction of human complex...

Area of Science:

• Genetics
• Genomics
• Computational Biology

Background:

• Human genetic association studies are crucial for understanding complex traits.
• Traditional methods using single-nucleotide polymorphisms (SNPs) have limitations in capturing functional genomic information.
• Identifying causal variants for complex diseases remains a challenge.

Purpose of the Study:

• To introduce a novel framework, Haplotype Function Score (HFS), for human genetic association studies.
• To leverage deep learning for calculating functional genomic activity scores from haplotypes.
• To improve the identification of genetic associations and causal variants for complex traits.

Main Methods:

• Developed a deep learning model (Sei) to compute Haplotype Function Scores (HFS) for individual haplotypes.
• Applied the HFS framework to analyze 14 complex traits in the UK Biobank data.
• Utilized fine-mapping and enrichment analyses to identify causal associations and biological pathways.
• Integrated HFS with SNP-based polygenic risk scores using LASSO regression.

Main Results:

• Identified 3619 significant HFS-trait associations (p < 5 × 10^-8).
• Fine-mapping revealed 2699 causal associations, a median increase of 63 per trait compared to SNP analysis.
• Uncovered 727 pathway-trait and 153 tissue-trait associations with biological interpretability.
• Achieved 16.1-39.8% improvement in cross-ancestry polygenic prediction by integrating HFS.

Conclusions:

• Haplotype Function Score (HFS) offers a powerful new strategy for human genetic association studies.
• The HFS framework significantly enhances the detection of causal genetic variants and biological insights.
• This approach holds promise for advancing our understanding of the genetic architecture of complex human traits.