A Chromosome-Scale Assembly of the Cashew Tree Genome: Leveraging the Combined Power of PacBio, DNA-Chromatin Proximity Ligation and SNP Linkage Maps

Widely cultivated in the tropical world for nuts and apples, the cashew tree (Anacardium occidentale) is still a genomics-orphan fruit crop. To unlock its genetic diversity and accelerate breeding, we have generated a chromosome-scale assembly of the ~488 Mbp genome of the dwarf variety BRS226 and an Axiom 16KSNP-chip from whole-genome resequence data of 24 diverse cashew accessions. A total of 27 Gb of sequence data (3,082,347 reads; N50: 12,142 bp; coverage: ~54x) from 41 PacBio RS-II/P6-C4 chemistry SMRT cells, was assembled into 2,577 contigs, covering 452Mb (92.6% of the genome; NG50/LG50: 1.05 Mb/113; longest sequence: 7.3 Mb) using Falcon Assembler. Filtering contigs deemed to contain locally alternative alleles not fused into a single haplotype representation of the heterozygous genome, resulted in an initial assembly of 377Mb (77% of the genome). Assembly contiguity was 4x improved (N50/L50: 1.5Mb/74 to N50/L50: 4.3Mb/30) following a HiRise scaffolding by Dovetail-Genomics using a 508.4x physical coverage ChicagoTM library. This improved assembly was anchored to eight single-tree linkage maps built from 359 individuals across five full-sib families encompassing 5,857 uniquely mapped SNPs, placing 352Mb (72% of the genome) into the 21 expected chromosomes. Concurrently, a DNA-chromatin proximity ligation Hi-C library was also used to improve the initial assembly contiguity, likewise closely converging to the 21 expected chromosomes (N90/L90: 11Mb/20). Additional genome coverage is being pursued by assembling the PacBio data into the correctly phased diploid genome by combining long and short reads with the SNP data. RNAseq from multiple tissues is underway for genome annotation.