Construction of genus-wide core genome
To construct the core-genome of Vitis genus, nine genomes were aligned to the 12X.2 versions of the PN40024 reference genome \cite{Jaillon2007} to identify the core genome. Due to the fact the 41.4% of the grape genome is composed of transposable element, we masked the repetitive genome regions before the alignment. The pairwise syntenic alignments are further filtered to ensure one-to-one match between the assembly and PN40024, as shown in Figure 1 A. On average, total length for the one-to-one match ranges from 160 Mb to 226Mb, depending on their genetic distance to V.vinifera. The collinearity between the assembly and the reference at genome level are high, which indicates that the overall quality of the assembly is good. The assembly has even coverage in both the arm and pericentromeric regions, but the scaffolds are smaller and less continuous in the pericentromeric regions than those in the arms, which indicates the pericentromeric regions are still challenging to assemble when using the 10X de novo genomes technology. As expected, the wild species represent a larger genetic divergence to the reference than that of cultivated modern cultivars. For example, on chromosome 9, on average 66% of the chromosome is collinear with wild species, while the percentage on average increases to 88% with cultivars. (Figure 1C, supplement table 1]. We defined the coverage of a chromosome region as how many times it occurs in a collinear alignment between the genome assemblies and the reference PN40024. About 10 % of the reference genome was covered by all of the genome assemblies, and these regions contain 64% of the annotated genes in total. This 64% is defined as core genes. This percentage is similar in rice, which is 62%, according to a recent study of 67 rice genomes.