Pan-NLRome profiling of NLR neighborhoods reveals extensive diversity in melon and boosts GWAS-based resistance gene discovery

Abstract Nucleotide-binding domain leucine-rich repeat genes (NLRs) play a key role in plant immunity. Although the main mechanisms driving NLR diversification have been outlined, the allelic diversity and genomic frameworks in which NLRs evolve have been poorly characterized, limiting the discovery of resistance-associated variants. We targeted all NLR neighborhoods across 143 melon accessions using long-read sequencing, the largest set in any species up to date, allowing accurate assemblies and gene annotation. Globally, we observed 21 NLR regions that were positionally conserved across melon genomes, yet highly variable in size, NLR content, and sequence complexity. We captured these variations in a comprehensive pan-NLRome graph. We assessed NLR allelic diversity using saturation curves and the Chao2 estimator, revealing unexpected allelic variation peaking in melon domestication centers and overcoming the tunnel vision derived from orthogroup-based analyses. Building on the assembled pan-NLRome, we proposed two genome-wide association study (GWAS) pipelines based on variation patterns of pan-NLRome graph nodes and k-mers. We benchmarked these methods on phenotypic data for three diseases, accurately identifying two previously validated NLR genes and their alleles, and identified additional novel NLR and non-NLR candidate genes. We established a high-resolution pan-NLRome for melon, revealing an extensive allelic diversity potentially explaining how Cucurbitaceae species withstand strong pathogen pressure despite their reduced NLR repertory. By integrating pan-NLRome diversity within GWAS, we established a powerful framework for resistance gene discovery beyond the limits of single-reference approaches. This strategy is portable to other diseases and crops, supporting the development of varieties adapted to current agroecological challenges.