Fig. 1

Overview of the workflow. a The Altus genome was sequenced using PacBio HiFi technology, whereas the 193 genomes of the cross Altus × Colomba were sequenced on the Illumina platform. b We used hifiasm to assemble the Altus HiFi reads into an assembly graph. For each contig in the graph, unique k-mers were detected (denoted by the colored bars). c The HiFi reads were aligned to the contigs and the mapping depth was used to estimate dosages (1 to 4) for each contig. The different dosages are denoted by the thickness of the contig line (thicker outlines mean higher dosage). d The unique k-mers were counted in the short reads of the offspring samples in order to compose a count pattern for each contig. e For all nodes from the assembly graph components, the pairwise correlation of k-mer count patterns was computed and components were clustered to represent chromosomes. f In each chromosome cluster, the nodes with estimated dosage 1 were first clustered into the four haplotypes, again based on pairwise correlations. g The contigs with dosages > 1 were added to the clusters that contain the most matching nodes in terms of k-mer count pattern correlations. h This process resulted in chromosome clusters that contain subclusters for each haplotype