SciGroveBeta
Genetics

HaploPerturb: Low-rank copula construction of haplotype perturbations improves sequence-to-function analysis of Alzheimer's disease loci

Jichun Xie

Featured August 21, 2026

Original Paper

AI-generated analysis — This is SciGrove's AI interpretation of the paper, not peer-reviewed content. Always refer to the original paper.

Simply

Generating realistic genetic inputs for disease prediction is tricky because DNA changes often come in linked packages, not one by one. This paper's HaploPerturb method creates these realistic packages, improving how we find disease-related genetic effects.

In depth
The paper introduces HaploPerturb, a novel framework that addresses the problem of generating realistic genetic sequence inputs for sequence-to-function models. Instead of perturbing only a single variant, which can create biologically implausible haplotypes, HaploPerturb uses a fixed-margin latent Gaussian factor model to capture linkage disequilibrium (LD) among variants. This allows the construction and ranking of high-probability, population-informed haplotype configurations conditional on a lead variant's allele, significantly improving the identification of cell-type-specific eQTLs in Alzheimer's disease loci.

Key Takeaways

  • 1
    Existing sequence-to-function models often use single-variant perturbations, which implicitly fix linked alleles to a reference state, potentially creating uncommon or unobserved population haplotypes.
  • 2
    The HaploPerturb framework models linkage disequilibrium (LD) using a latent Gaussian factor model to construct and rank population-informed haplotypes, ensuring more realistic genetic inputs for functional prediction.
  • 3
    Applying HaploPerturb to Alzheimer's disease loci significantly improves the microglial eQTL ranking compared to single-lead edits, demonstrating the importance of haplotype-aware input construction for functional interpretation.

Conceptual Flow

HIGH LEVEL
1
Methodology: Creating Realistic DNA Inputs

Instead of changing just one DNA spot, the new method looks at how many spots usually change together to make a more natural-looking DNA sequence.

Lead DNA Change
Nearby DNA Changes
Find Common Patterns
Realistic DNA Sequence
2
Results: Better Disease Gene Discovery

By using these more natural DNA sequences, the method got much better at finding which DNA changes affect brain cell activity in Alzheimer's disease.

Old DNA Sequences
New DNA Sequences
Compare Disease Links
Clearer Disease Genes

This breakdown was generated by SciGrove. Get the same analysis — intuition, storyboard, peer review, a runnable prototype and a glossary — on any paper you upload or paste a DOI for.