Deciphering the mosaic genome of sugarcane cultivars through polyploid admixture inference with AdmixPoly

Characterizing population structure and admixture events between ancestral groups plays a key role in understanding the evolutionary history of species and crops. Most tools for inferring admixture have been developed for diploids and are not suitable for polyploids, in particular those with high and mixed ploidy such as Saccharum.

Here we present AdmixPoly, an R-package designed to infer admixture in polyploid species both at the genome-wide scale and locally along chromosomes. We compare AdmixPoly with state-of-the-art methods using simulations, demonstrating its precision and computational efficiency. Notably, local admixture inference in complex scenarios, such as high ploidy levels, large numbers of ancestral groups and alleles per marker is enabled through efficient approximations of emission and transition probabilities within a hidden Markov model framework. We apply this approach to characterize the contributions of wild Saccharum species to the complex polyploid genome of modern sugarcane cultivars. A panel of wild and cultivated Saccharum accessions is genotyped for 80K genomic regions, each revealing approximately 50 read-scale haplotypes.

The results reveal that most of the approximately 12 copies of each basic chromosome in modern cultivars are derived from the domesticated species Saccharum officinarum, with one to four copies typically contributed by distinct subgroups of the wild species Saccharum spontaneum. In addition, contributions from an unknown wild Saccharum group originating from the Pacific were identified in most cultivars. The conserved pattern of these introgressions suggests that they can be traced back to the early stages of sugarcane breeding approximately a century ago.