Cassava genomes assembled with single-molecule long reads, optical and Hi-C maps reveal narrow genetic diversity and mono-allelic expression

Cassava (Manihot Esculenta Crantz) is an important food security crop for nearly one-billion people in tropical and sub-tropical regions worldwide. But genetic improvement of cassava is constrained by the proportion of deleterious mutations in coding sequences and highly fragmented, incomplete draft genome assemblies 1–3. Full cassava genome assemblies have not been achieved because of the excessive heterozygous genetic composition and diploid status of the genome. Here we present the first diploid-aware assemblies and annotation of genomes for two African cassava varieties (TME 3 and 60444) using single-molecule real-time sequencing, combined with high-resolution optical mapping and chromosome proximity ligation data to create chromosomal sequence scaffolds. We revised and improved the cassava de novo predicted gene space using full-length, single-molecule CDS sequencing and analysed the transcriptome for allele-specific expression. The two high-quality cassava genomes have a near 1.3 Gb diploid genome size, reveal the repetitive DNA proportion in detail, and phase thousands of allelic variants in mega-base-pair haplotype blocks. We expect that the high-quality genomes will facilitate targeted molecular breeding and gene isolation to improve cassava.