SUMMARY In order to provide a comprehensive resource for human structural variants (SVs), we generated long-read sequence data and analyzed SVs for fifteen human genomes. We sequence resolved 99,604 insertions, deletions, and inversions including 2,238 (1.6 Mbp) that are shared among all discovery genomes with an additional 13,053 (6.9 Mbp) present in the majority, indicating minor alleles or errors in the reference. Genotyping in 440 additional genomes confirms the most common SVs in unique euchromatin are now sequence resolved. We report a ninefold SV bias toward the last 5 Mbp of human chromosomes with nearly 55% of all VNTRs (variable number of tandem repeats) mapping to this portion of the genome. We identify SVs affecting coding and noncoding regulatory loci improving annotation and interpretation of functional variation. These data provide the framework to construct a canonical human reference and a resource for developing advanced representations capable of capturing allelic diversity.

Genetic studies of human evolution require high-quality contiguous ape genome assemblies that are not guided by the human reference. We coupled long-read sequence assembly, full-length cDNA sequencing with a multi-platform scaffolding approach to produce ab initio chimpanzee and orangutan genome assemblies. Comparing these with two long-read de novo human genome assemblies and a gorilla genome assembly, we characterized lineage-specific and shared great ape genetic variation ranging from single base-pair to megabase-sized variants. We identified ~17 thousand fixed human-specific structural variants identifying genic and putative regulatory changes that emerged in humans since divergence from nonhuman apes. Interestingly, these fixed human-specific structural variants are enriched near genes that are downregulated in human compared to chimpanzee cerebral organoids, particularly in cells analogous to radial glial neural progenitors.

De novo assembly of a human genome using nanopore long-read sequences has been reported but it used more than 150,000 CPU hours and weeks of wall-clock time. To enable rapid human genome assembly we present Shasta, a de novo long read assembler, and polishing algorithms named MarginPolish and HELEN. Using a single PromethION nanopore sequencer and our toolkit, we assembled eleven highly contiguous human genomes de novo in nine days. We achieved ~63x coverage, 42 Kb read N50, and 6.5x coverage in 100 Kb+ reads using three flow cells per sample. Shasta produced a complete haploid human genome assembly in under six hours on a single commercial compute node. MarginPolish and HELEN polished haploid assemblies to more than 99.9% identity (QV30) with nanopore reads alone. Addition of proximity ligation (Hi-C) sequencing enabled near chromosome-level scaffolds for all eleven genomes. We compare our assembly performance to existing methods for diploid, haploid, and trio-binned human samples and report superior accuracy and speed.

The complete assembly of each human chromosome is essential for understanding human biology and evolution1,2. Here we use complementary long-read sequencing technologies to complete the linear assembly of human chromosome 8. Our assembly resolves the sequence of five previously long-standing gaps, including a 2.08-Mb centromeric α-satellite array, a 644-kb copy number polymorphism in the β-defensin gene cluster that is important for disease risk, and an 863-kb variable number tandem repeat at chromosome 8q21.2 that can function as a neocentromere. We show that the centromeric α-satellite array is generally methylated except for a 73-kb hypomethylated region of diverse higher-order α-satellites enriched with CENP-A nucleosomes, consistent with the location of the kinetochore. In addition, we confirm the overall organization and methylation pattern of the centromere in a diploid human genome. Using a dual long-read sequencing approach, we complete high-quality draft assemblies of the orthologous centromere from chromosome 8 in chimpanzee, orangutan and macaque to reconstruct its evolutionary history. Comparative and phylogenetic analyses show that the higher-order α-satellite structure evolved in the great ape ancestor with a layered symmetry, in which more ancient higher-order repeats locate peripherally to monomeric α-satellites. We estimate that the mutation rate of centromeric satellite DNA is accelerated by more than 2.2-fold compared to the unique portions of the genome, and this acceleration extends into the flanking sequence.

We developed a computational method based on polyploid phasing of long sequence reads to resolve collapsed regions of segmental duplications within genome assemblies. The approach, Segmental Duplication Assembler (SDA), constructs graphs where paralogous sequence variants define the nodes and long-read sequences provide attraction and repulsion edges allowing us to partition and assemble long reads corresponding to distinct paralogs. We apply it to single-molecule, real-time sequence data from three human genomes and recover 33–79 Mbp of duplications where approximately half of the loci are diverged (<99.8%) when compared to the reference genome. We show that the corresponding sequence is highly accurate (>99.9%) and that the diverged sequence corresponds to copy number variable paralogs that are absent from the human reference. Our method can be applied to other complex genomes to resolve the last gene-rich gaps, improve duplicate gene annotation, and better understand copy number variant genetic diversity at the base-pair level.

Human genomes are typically assembled as consensus sequences that lack information on parental haplotypes. Here we describe a reference-free workflow for diploid de novo genome assembly that combines the chromosome-wide phasing and scaffolding capabilities of single-cell strand sequencing1,2 with continuous long-read or high-fidelity3 sequencing data. Employing this strategy, we produced a completely phased de novo genome assembly for each haplotype of an individual of Puerto Rican descent (HG00733) in the absence of parental data. The assemblies are accurate (quality value > 40) and highly contiguous (contig N50 > 23 Mbp) with low switch error rates (0.17%), providing fully phased single-nucleotide variants, indels and structural variants. A comparison of Oxford Nanopore Technologies and Pacific Biosciences phased assemblies identified 154 regions that are preferential sites of contig breaks, irrespective of sequencing technology or phasing algorithms.

Copy number variants (CNVs) are subject to stronger selective pressure than single-nucleotide variants, but their roles in archaic introgression and adaptation have not been systematically investigated. We show that stratified CNVs are significantly associated with signatures of positive selection in Melanesians and provide evidence for adaptive introgression of large CNVs at chromosomes 16p11.2 and 8p21.3 from Denisovans and Neanderthals, respectively. Using long-read sequence data, we reconstruct the structure and complex evolutionary history of these polymorphisms and show that both encode positively selected genes absent from most human populations. Our results collectively suggest that large CNVs originating in archaic hominins and introgressed into modern humans have played an important role in local population adaptation and represent an insufficiently studied source of large-scale genetic variation.

Present workflows for producing human genome assemblies from long-read technologies have cost and production time bottlenecks that prohibit efficient scaling to large cohorts. We demonstrate an optimized PromethION nanopore sequencing method for eleven human genomes. The sequencing, performed on one machine in nine days, achieved an average 63x coverage, 42 Kb read N50, 90% median read identity and 6.5x coverage in 100 Kb+ reads using just three flow cells per sample. To assemble these data we introduce new computational tools: Shasta -a de novo long read assembler, and MarginPolish & HELENa suite of nanopore assembly polishing algorithms. On a single commercial compute node Shasta can produce a complete human genome assembly in under six hours, and MarginPolish & HELEN can polish the result in just over a day, achieving 99.9% identity (QV30) for haploid samples from nanopore reads alone. We evaluate assembly performance for diploid, haploid and trio-binned human samples in terms of accuracy, cost, and time and demonstrate improvements relative to current state-of-the-art methods in all areas. We further show that addition of proximity ligation (Hi-C) sequencing yields near chromosome-level scaffolds for all eleven genomes.