Skip to content

AI GWAS Review

Curation of Review within GWAS — listings under the AI tab.

AI-enhanced GWAS

The GWAS topic within AI covers methods that use machine learning to boost genome-wide association studies. Main trajectories:

  • Imaging GWAS: Early work applied supervised CNNs (U-Net, ResNet) to medical images for trait quantification and GWAS (Haas PMID 34957434, Cell Genomics 2021; Khurshid PMID 36944631, Nat Commun 2023). Recent work uses self-supervised contrastive learning to bypass manual annotation, running GWAS directly on image embeddings (iGWAS, Kirchler et al. PMID 39020183, Nature Genetics 2024).

  • AI Phenotyping: Ensemble ML and multimodal topic models (EHR + genetics) for phenotype extraction and GWAS on imputed phenotypes (MILTON, Garg et al. PMID 39471869, Nat Genet 2024; MixEHR-SAGE, Cui et al. PMID 39843619, Nat Med 2026).

  • AI Association Methods: Neural network-based association tests (GWANN, Holzinger et al. PMID 38918402, Nat Genet 2024) and Transformer-based models for discovering novel loci (InsightGWAS, Song et al. bioRxiv 2025).

  • Post-GWAS AI: Gene prioritization using polygenic features (PoPS, Weeks et al. PMID 37106029, Nat Genet 2023).

  • Methodology reviews: Causal ML for single-cell genomics (Tejada-Lapuerta et al. PMID 40336376, Nat Genet 2025).

Summary Table

Click a column header to sort the table.

NAME Main citation YEAR
Causal ML for scGenomics
Tejada-Lapuerta A et al., Nat Genet, 2025
 2025
DL for PRS Survey
Schuran M et al., Brief Bioinform, 2025
 2025
Imaging Genomics Review
Bian Y et al., Nat Rev Genet, 2026
 2026

Causal ML for scGenomics (Causal ML sc)

AI GWAS Causal ML Single Cell Machine Learning Nat Genet
PUBMED_LINK
40164735
FULL NAME
Causal Machine Learning for Single-Cell Genomics
DESCRIPTION
A Perspective from Nature Genetics delineating the application of causal machine learning to single-cell genomics. Discusses causal models, challenges in inferring causative roles of genes from single-cell omics data combined with perturbation screens, and the potential for integrating causal ML with GWAS to understand disease mechanisms at single-cell resolution.
TITLE
Causal machine learning for single-cell genomics.
ABSTRACT
Advances in single-cell '-omics' allow unprecedented insights into the transcriptional profiles of individual cells and, when combined with large-scale perturbation screens, enable measuring of the effect of targeted perturbations on the whole transcriptome. In this Perspective, we delineate the application of causal machine learning to single-cell genomics and its associated challenges, presenting the causal model most commonly applied to single-cell biology.
DOI
10.1038/s41588-025-02124-2

DL for PRS Survey (DL PRS Survey)

AI GWAS Polygenic Risk Score Deep Learning Survey Review Brief Bioinform
PUBMED_LINK
40802796
FULL NAME
A Survey on Deep Learning for Polygenic Risk Scores
DESCRIPTION
A comprehensive survey of deep learning approaches for polygenic risk scores (PRS). Reviews how neural networks can model non-linear relationships between genetic variants and disease risk, going beyond traditional linear PRS methods, and assesses their performance across different traits and architectures. Published in Briefings in Bioinformatics.
TITLE
A survey on deep learning for polygenic risk scores.
ABSTRACT
Polygenic risk scores (PRS) combine the effects of multiple genetic variants to predict an individual's genetic predisposition to a disease. PRS typically rely on linear models, which assume that all genetic variants act independently. There is growing interest in applying deep learning neural networks to model PRS given their ability to model non-linear relationships. We conducted a survey of the literature to investigate how neural networks are being applied to PRS.
DOI
10.1093/bib/bbaf373

Imaging Genomics Review

AI GWAS Imaging Imaging-derived phenotypes Review IDP MRI Deep Learning Mendelian Randomization Nat Rev Genet
PUBMED_LINK
42409965
FULL NAME
Genetic analysis of imaging-derived phenotypes
DESCRIPTION
A comprehensive review of imaging-derived phenotypes (IDPs) for genetic analysis. Covers MRI, CT, X-ray, OCT, and other imaging modalities; traditional (FSL, FreeSurfer) and deep learning (U-Net, nnU-Net, self-supervised contrastive learning) pipelines for IDP extraction; biobank-scale cohorts (UK Biobank, All of Us); GWAS of brain, cardiac, retinal, abdominal, and skeletal IDPs; Mendelian randomization for causal inference linking organ structure to disease; and emerging challenges in multi-organ phenomics, longitudinal imaging, and clinical translation of IDP-based PRS.
URL
https://www.nature.com/articles/s41576-026-00989-5
TITLE
Genetic analysis of imaging-derived phenotypes.
Main citation
Bian Y, Akey JM. (2026) Genetic analysis of imaging-derived phenotypes. Nature Reviews Genetics. doi:10.1038/s41576-026-00989-5. PMID 42409965
ABSTRACT
Imaging-derived phenotypes (IDPs) developed from medical imaging data, such as magnetic resonance imaging, computed tomography and X-ray scans, are traits that provide quantitative information on anatomical and functional properties of organs and tissues. IDPs are powerful tools for identifying biomarkers and studying disease mechanisms. When coupled with genetic data, IDPs can be analysed as heritable phenotypes using modern gene mapping methods to uncover genotype-phenotype relationships. The field of imaging genomics is rapidly maturing, with the emergence of high-quality imaging datasets collected in biobank-scale cohorts and sophisticated computational methods for extracting IDPs from imaging data, including tools that leverage machine learning. Here we review common imaging modalities and analytical approaches for developing IDPs, discuss biological insights gleaned from the large-scale genetic analysis of imaging traits and highlight emerging areas and remaining challenges that must be overcome to realize the full potential of IDPs for genetic analysis.
DOI
10.1038/s41576-026-00989-5