Revisiting tandem duplication in plant genomes: Technical challenges and pangenome solutions

Tandem arrays, genomic loci comprising adjacent paralogs that share high sequence identity, concentrate in plant genome regions shaped by strong adaptive pressure, from nucleotide-binding leucine-rich repeat (NLR) resistance clusters to specialized metabolism loci. Yet these biologically informative neighborhoods are precarious bioinformatically. Short-read assemblies often result in assembly collapse, compressing multi-member tandem arrays into a single consensus sequence. Annotation pipelines compound this error through annotation fusion, which merges distinct array members into elongated gene models, and annotation omission, which drops true array members even when assemblies preserve local structure. The result is systemic distortion: collapsed references misrepresent tandem array copy number, confounded expression quantification across array members, and obscured tandem-array copy number variation (CNV) in population-genomic analyses. Graph pangenomes built from long-read, haplotype-resolved assemblies offer a direct remedy. By representing alternative locus structures as paths in a pangenome graph, these references restore tandem arrays as discoverable, measurable objects. Individual array members retain distinct coordinates, enabling array-member-resolved expression analysis and accurate genotyping of tandem-array CNV. This shift turns tandem-array-rich loci from systemic blind spots into accessible windows on adaptive genome evolution.