Pangenome Graph Node-Phenotype Association shows GWAS-like quality results with only few individuals
- 1IRD - French institute for Sustainable Development
Abstract
Purpose: We introduce GraNPA (Graph Node-Phenotype Association), a method that performs a phenotype-sequence association analysis on a pangenome vari- ation graph (PVG). This PVG is built using a small number of individual genome sequences and does not require additional population materials or kinship information for qualitative phenotypes. This approach minimizes the number of individuals required for association studies and mitigates the impact of reference bias in variant calling within these types of analyses. Background: A PVG is a representation of the multiple alignment of a set of complete genomes. It contains all variations, ranging from single nucleotide poly- morphisms (SNPs) to large structural variations (SVs), and these are represented as nodes in the graph. By integrating phenotype information within these nodes, we can assign a Score (S ) to each one and identify regions related to the phe- notype directly within the PVG. These regions represent statistically significant shifts in S distribution, indicating their involvement in the phenotype. Finally, GraNPA provides their positions and scores for further analysis. Results: This method was tested using different simulated contexts and two publicly available datasets: the Sub1A gene locus for Asian rice Oryza sativa in a 13 individuals PVG, and the insertion responsible for the white-headed in cattle (Bos taurus) with a PVG of 24 individuals. Source code of GraNPA is available here https://forge.ird.fr/diade/graphgwas/granpa under GNU GPLv3. Conclusion: GraNPA was able to identify the expected area in two simulated datasets, as well as the loci responsible for these two known traits, using only a few dozen complete genomes from these PVGs. Although this method is currently limited to qualitative phenotypes, it paves the way for more efficient methods that rely on PVGs and a small number of individuals.
Related
- Genotype-phenotype analysis in 333 individuals with heterozygous variants in NFKB1.
- A bacterial pangenome association study identifies a role for polyamine metabolism and alkaline tolerance in middle ear infections
- Association of calcium phosphate crystals with pain and inflammation among individuals with advanced knee osteoarthritis: A cross-sectional study.
- Change in hearing loss and its association with general health in a community-based longitudinal study of middle-aged individuals: The Busselton healthy ageing study
- Mirikizumab is associated with improvements in bowel urgency and quality of life in moderately-to-severely active ulcerative colitis: 28-week phase 3b results from the LUCENT-URGE trial.
BibTeX
@misc{carrette2026pangenome,
title = {{Pangenome Graph Node-Phenotype Association shows GWAS-like quality results with only few individuals}},
author = {C. Carrette and F. Sabot and C. Muller},
howpublished = {bioRxiv},
year = {2026},
month = jul,
doi = {10.64898/2026.07.31.741971},
url = {https://www.biorxiv.org/content/10.64898/2026.07.31.741971v2},
note = {CC BY},
}
RIS
TY - UNPB
TI - Pangenome Graph Node-Phenotype Association shows GWAS-like quality results with only few individuals
AU - C. Carrette
AU - F. Sabot
AU - C. Muller
PY - 2026
DA - 2026/07/31
PB - bioRxiv
DO - 10.64898/2026.07.31.741971
UR - https://www.biorxiv.org/content/10.64898/2026.07.31.741971v2
AB - Purpose: We introduce GraNPA (Graph Node-Phenotype Association), a method that performs a phenotype-sequence association analysis on a pangenome vari- ation graph (PVG). This PVG is built using a small number of individual genome sequences and does not require additional population materials or kinship information for qualitative phenotypes. This approach minimizes the number of individuals required for association studies and mitigates the impact of reference bias in variant calling within these types of analyses. Background: A PVG is a representation of the multiple alignment of a set of complete genomes. It contains all variations, ranging from single nucleotide poly- morphisms (SNPs) to large structural variations (SVs), and these are represented as nodes in the graph. By integrating phenotype information within these nodes, we can assign a Score (S ) to each one and identify regions related to the phe- notype directly within the PVG. These regions represent statistically significant shifts in S distribution, indicating their involvement in the phenotype. Finally, GraNPA provides their positions and scores for further analysis. Results: This method was tested using different simulated contexts and two publicly available datasets: the Sub1A gene locus for Asian rice Oryza sativa in a 13 individuals PVG, and the insertion responsible for the white-headed in cattle (Bos taurus) with a PVG of 24 individuals. Source code of GraNPA is available here https://forge.ird.fr/diade/graphgwas/granpa under GNU GPLv3. Conclusion: GraNPA was able to identify the expected area in two simulated datasets, as well as the loci responsible for these two known traits, using only a few dozen complete genomes from these PVGs. Although this method is currently limited to qualitative phenotypes, it paves the way for more efficient methods that rely on PVGs and a small number of individuals.
ER -