CoChat

Nanopore sequencing and the Shasta toolkit enable efficient de novo assembly of eleven human genomes

Open access

Published version of a preprint

The DOI matches its Crossref record; 78 of the 78 references it cites were found in a registry or an index.

Abstract

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 11 highly contiguous human genomes de novo in 9 d. We achieved roughly 63× coverage, 42-kb read N50 values and 6.5× coverage in reads >100 kb using three flow cells per sample. Shasta produced a complete haploid human genome assembly in under 6 h on a single commercial compute node. MarginPolish and HELEN polished haploid assemblies to more than 99.9% identity (Phred quality score QV = 30) with nanopore reads alone. Addition of proximity-ligation sequencing enabled near chromosome-level scaffolds for all 11 genomes. We compare our assembly performance to existing methods for diploid, haploid and trio-binned human samples and report superior accuracy and speed.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/s41587-020-0503-6

Cites

78 of 78 references verified0 citation errors · 0 not checkable

Statistics

Figures

Fig. 1 Nanopore sequencing data. a, Throughput in gigabases from each of three flow cells for 11 samples, with total throughput at top. Each point is a flow cell. b, Read N50 values for each flow cell. Each point is a flow cell. c, Alignment identities against GRCh38. Medians in a–c shown by dashed lines, dotted line in c is the mode. Each line is a single sample comprising three flow cells. d, Genome coverage as a function of read length. Dashed lines indicate coverage at 10 and 100 kb. HG00733 is accentuated in dark blue as an example. Each line is a single sample comprising three flow cells. e, Alignment identity for standard and RLE reads. Data for HG00733 chromosome 1 flow cell 1 are shown (4.6 Gb raw sequence). Dashed lines denote quartiles. Source data
Fig. 1 Nanopore sequencing data. a, Throughput in gigabases from each of three flow cells for 11 samples, with total throughput at top. Each point is a flow cell. b, Read N50 values for each flow cell. Each point is a flow cell. c, Alignment identities against GRCh38. Medians in a–c shown by dashed lines, dotted line in c is the mode. Each line is a single sample comprising three flow cells. d, Genome coverage as a function of read length. Dashed lines indicate coverage at 10 and 100 kb. HG00733 is accentuated in dark blue as an example. Each line is a single sample comprising three flow cells. e, Alignment identity for standard and RLE reads. Data for HG00733 chromosome 1 flow cell 1 are shown (4.6 Gb raw sequence). Dashed lines denote quartiles. Source dataCC BY · PubMed Central
Fig. 2 Assembly metrics for Shasta, Wtgdb2, Flye and Canu before polishing. a, NGx plot showing contig length distribution. The intersection of each line with the dashed line is the NG50 for that assembly. b, NGAx plot showing the distribution of aligned contig lengths. Each horizontal line represents an aligned segment of the assembly unbroken by a disagreement or unmappable sequence with respect to GRCh38. The intersection of each line with the dashed line is the aligned NGA50 for that assembly. c, Assembly disagreement counts for regions outside centromeres, segmental duplications and, for HG002, known SVs. d, Total generated sequence length versus total aligned sequence length (against GRCh38). e, Balanced base-level error rates for assembled sequences. f, Average runtime and cost for assemblers (Canu not shown). Source data
Fig. 2 Assembly metrics for Shasta, Wtgdb2, Flye and Canu before polishing. a, NGx plot showing contig length distribution. The intersection of each line with the dashed line is the NG50 for that assembly. b, NGAx plot showing the distribution of aligned contig lengths. Each horizontal line represents an aligned segment of the assembly unbroken by a disagreement or unmappable sequence with respect to GRCh38. The intersection of each line with the dashed line is the aligned NGA50 for that assembly. c, Assembly disagreement counts for regions outside centromeres, segmental duplications and, for HG002, known SVs. d, Total generated sequence length versus total aligned sequence length (against GRCh38). e, Balanced base-level error rates for assembled sequences. f, Average runtime and cost for assemblers (Canu not shown). Source dataCC BY · PubMed Central
Fig. 3 Shasta MHC assemblies compared with the reference human genome. Unpolished Shasta assembly for CHM13 and HG00733, including HG00733 trio-binned maternal and paternal assemblies. Shaded gray areas are regions in which coverage (as aligned to GRCh38) drops below 20. Horizontal black lines indicate contig breaks. Blue and green describe unique alignments (aligning forward and reverse, respectively) and orange describes multiple alignments. Source data
Fig. 3 Shasta MHC assemblies compared with the reference human genome. Unpolished Shasta assembly for CHM13 and HG00733, including HG00733 trio-binned maternal and paternal assemblies. Shaded gray areas are regions in which coverage (as aligned to GRCh38) drops below 20. Horizontal black lines indicate contig breaks. Blue and green describe unique alignments (aligning forward and reverse, respectively) and orange describes multiple alignments. Source dataCC BY · PubMed Central
Fig. 4 Polishing assembled genomes. a, Balanced error rates for the four methods on HG00733 and CHM13. b, Row-normalized heatmaps describing the predicted run lengths (x axis) given true run lengths (y axis) for four steps of the pipeline on HG00733. Guppy v.2.3.3 was generated from 3.7 Gb of RLE sequence. Shasta, MarginPolish and HELEN were generated from whole assemblies aligned to their respective truth sequences. c, Error rates for MarginPolish and HELEN on four assemblies. d, Average runtime and cost. Source data
Fig. 4 Polishing assembled genomes. a, Balanced error rates for the four methods on HG00733 and CHM13. b, Row-normalized heatmaps describing the predicted run lengths (x axis) given true run lengths (y axis) for four steps of the pipeline on HG00733. Guppy v.2.3.3 was generated from 3.7 Gb of RLE sequence. Shasta, MarginPolish and HELEN were generated from whole assemblies aligned to their respective truth sequences. c, Error rates for MarginPolish and HELEN on four assemblies. d, Average runtime and cost. Source dataCC BY · PubMed Central
Fig. 5 HiRise scaffolding for 11 genomes. a, NGx plots for each of the 11 genomes, before (dashed) and after (solid) scaffolding with HiC sequencing reads, GRCh38 minus alternate sequences is shown for comparison. b, Dot plot showing alignments between the scaffolded HG00733 Shasta assembly and GRCh38 chromosome scaffolds. Blue indicates forward aligning segments, green indicates reverse, with both indicating unique alignments. Source data
Fig. 5 HiRise scaffolding for 11 genomes. a, NGx plots for each of the 11 genomes, before (dashed) and after (solid) scaffolding with HiC sequencing reads, GRCh38 minus alternate sequences is shown for comparison. b, Dot plot showing alignments between the scaffolded HG00733 Shasta assembly and GRCh38 chromosome scaffolds. Blue indicates forward aligning segments, green indicates reverse, with both indicating unique alignments. Source dataCC BY · PubMed Central
Extended Data Fig. 1 Read Markers. Markers aligned to a run length encoded read.
Extended Data Fig. 1 Read Markers. Markers aligned to a run length encoded read.CC BY · PubMed Central
Extended Data Fig. 2 Marker Alignment. A marker alignment represented as a dot-plot. Elements that are identical between the two sequences are displayed in green or red - the ones in green are the ones that are part of the optimal alignment computed by the Shasta assembler. Because of the much larger alphabet, matrix elements that are identical between the sequences but are not part of the optimal alignment are infrequent. Each alignment matrix element here corresponds on average to a 13 13 block in the alignment matrix in raw base sequence.
Extended Data Fig. 2 Marker Alignment. A marker alignment represented as a dot-plot. Elements that are identical between the two sequences are displayed in green or red - the ones in green are the ones that are part of the optimal alignment computed by the Shasta assembler. Because of the much larger alphabet, matrix elements that are identical between the sequences but are not part of the optimal alignment are infrequent. Each alignment matrix element here corresponds on average to a 13 13 block in the alignment matrix in raw base sequence.CC BY · PubMed Central
Extended Data Fig. 3 Read Graph. An example of a portion of the read graph (A) as displayed by the Shasta http server, and (B) showing obviously incorrect connections.
Extended Data Fig. 3 Read Graph. An example of a portion of the read graph (A) as displayed by the Shasta http server, and (B) showing obviously incorrect connections.CC BY · PubMed Central
Extended Data Fig. 4 Marker Graph. An illustration of marker graph construction for two sequences.
Extended Data Fig. 4 Marker Graph. An illustration of marker graph construction for two sequences.CC BY · PubMed Central
Extended Data Fig. 5 Assembly Graph. (A) A marker graph with linear sequence of edges colored. (B) The corresponding assembly graph. Colors were chosen to indicate the correspondence to marker graph edges.
Extended Data Fig. 5 Assembly Graph. (A) A marker graph with linear sequence of edges colored. (B) The corresponding assembly graph. Colors were chosen to indicate the correspondence to marker graph edges.CC BY · PubMed Central
Extended Data Fig. 6 Bubbles. (A) A simple bubble. (B) A superbubble.
Extended Data Fig. 6 Bubbles. (A) A simple bubble. (B) A superbubble.CC BY · PubMed Central
Extended Data Fig. 7 POA Example. (A) An example POA, assuming approximately 30x read coverage. The backbone is shown in red. Each non-source/sink node has a vector of weights, one for each possible base. Deletion edges are shown in teal, they also each have a weight. Finally insertion nodes are shown in brown, each also has a weight. (B) A pruned POA, removing deletions and insertions that have less than a threshold weight and highlighting plausible bases in bold. There are six plausible nucleotide sequences represented by paths through the POA and selections of plausible base labels: G;AT;A;T;A;C:A, G;AT;A;T;A;C:G, G;A;T;A;C:A, G;A;T;A;C:G, G;A;C:A, G;A;C:G. To avoid the combinatorial explosion of such enumeration we identify subgraphs (C) and locally enumerate the possible subsequences in these regions independently (dotted rectangles identify subgraphs selected). In each subgraph there is a source and sink node that does not overlap any proposed edit.
Extended Data Fig. 7 POA Example. (A) An example POA, assuming approximately 30x read coverage. The backbone is shown in red. Each non-source/sink node has a vector of weights, one for each possible base. Deletion edges are shown in teal, they also each have a weight. Finally insertion nodes are shown in brown, each also has a weight. (B) A pruned POA, removing deletions and insertions that have less than a threshold weight and highlighting plausible bases in bold. There are six plausible nucleotide sequences represented by paths through the POA and selections of plausible base labels: G;AT;A;T;A;C:A, G;AT;A;T;A;C:G, G;A;T;A;C:A, G;A;T;A;C:G, G;A;C:A, G;A;C:G. To avoid the combinatorial explosion of such enumeration we identify subgraphs (C) and locally enumerate the possible subsequences in these regions independently (dotted rectangles identify subgraphs selected). In each subgraph there is a source and sink node that does not overlap any proposed edit.CC BY · PubMed Central
Extended Data Fig. 8 RLE Inference Distributions. Visual representation of run length inference. This diagram shows how a consensus run length is inferred for a set of aligned lengths (X) that pertain to a single position. The lengths are factored and then iterated over, and log likelihood is calculated for every possible true length up to a predefined limit. Note that in this example, the most frequent observation (4bp) is not the most likely true length (5bp) given the model.
Extended Data Fig. 8 RLE Inference Distributions. Visual representation of run length inference. This diagram shows how a consensus run length is inferred for a set of aligned lengths (X) that pertain to a single position. The lengths are factored and then iterated over, and log likelihood is calculated for every possible true length up to a predefined limit. Note that in this example, the most frequent observation (4bp) is not the most likely true length (5bp) given the model.CC BY · PubMed Central
Extended Data Fig. 9 MarginPolish HELEN Image Generation. A graphical representation of images from two labeled regions selected to demonstrate: the encoding of a single POA node into two run-length blocks (i), a true deletion (i), and a true insert (ii). (a) shows the alignment in raw and run-length space, (b) shows the features as they are exported to HELEN. The y-axis shows truth labels for nucleotides and run-lengths, the x-axis describes features in the images, and colors show associated weights.
Extended Data Fig. 9 MarginPolish HELEN Image Generation. A graphical representation of images from two labeled regions selected to demonstrate: the encoding of a single POA node into two run-length blocks (i), a true deletion (i), and a true insert (ii). (a) shows the alignment in raw and run-length space, (b) shows the features as they are exported to HELEN. The y-axis shows truth labels for nucleotides and run-lengths, the x-axis describes features in the images, and colors show associated weights.CC BY · PubMed Central
Extended Data Fig. 10 HELEN Model. The sequence-to-sequence model implemented in Helen.
Extended Data Fig. 10 HELEN Model. The sequence-to-sequence model implemented in Helen.CC BY · PubMed Central

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 78 references

Research works printed but not linked: 1

  1. Vollger, M. R. et al. Improved assembly and variant detection of a haploid human genome using single-molecule, high-fidelity long reads. Ann. Hum. Genet. 84, 125–140 (2020). · DOI

Cited by

See all 570

Metadata

Authors
Shafin, K., Pesout, T., Lorig-Roach, R., Haukness, M., Olsen, H. E., Bosworth, C., Armstrong, J., Tigyi, K., Maurer, N., Koren, S., Sedlazeck, F. J., Marschall, T., Mayes, S., Costa, V., Zook, J. M., Liu, K. J., Kilburn, D., Sorensen, M., Munson, K. M., Vollger, M. R., Monlong, J., Garrison, E., Eichler, E. E., Salama, S., Haussler, D., Green, R. E., Akeson, M., Phillippy, A., Miga, K. H., Carnevali, P., Jain, M., & Paten, B.
Affiliations
  • University of California, Santa Cruz (US)
  • Howard Hughes Medical Institute (US)
  • National Institutes of Health (US)
  • Baylor College of Medicine (US)
  • Max Planck Institute for Informatics (DE)
  • Oxford Nanopore Technologies (United Kingdom) (GB)
  • National Institute of Standards and Technology (US)
  • Circulomics (United States) (US)
  • University of Washington (US)
  • Chan Zuckerberg Initiative (United States) (US)
Corresponding author
  • Carnevali, Paolo
  • Jain, Miten
  • Paten, Benedict
Venue
Nature Biotechnology
Published
2020-05-04
Volume
38
Issue
9
Pages
1044-1053
ISSN
  • 0733-222X (ISSN-L)
  • 1087-0156
  • 1546-1696
OpenAlex
W3022783334
Cited by
567
References
74
Citations by year
2026: 44 · 2025: 62 · 2024: 117 · 2023: 91 · 2022: 89 · 2021: 115 · 2020: 44
FWCI
27.90
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Biotechnology (publishedVersion, cc-by)
  • bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • HAL (submittedVersion)
  • eScholarship (California Digital Library) (submittedVersion)
  • MPG.PuRe (Max Planck Society) (submittedVersion, other-oa)
Funders
  • U.S. Department of Health & Human Services | NIH | National Heart, Lung, and Blood Institute (1U01HL137183)
  • U.S. Department of Health & Human Services | National Institutes of Health (5U54HG007990, 5T32HG008345-04, 1U01HL137183, R01HG010053)
  • U.S. Department of Health & Human Services | NIH | National Human Genome Research Institute (3U24HG009084-03S1, 2U41HG007234)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Genomics and Phylogenetic Studies
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Genomic variations and chromosomal abnormalities
  • Physical Sciences › Mathematics › Theoretical Computer Science › Algorithms and Data Compression
Keywords
nanopore sequencing, de novo genome assembly, Helen, human genome assembly, haploid genome assembly, long-read sequencing, chromosome-level scaffolding, assembly polishing
MeSH terms
Deep Learning, Nanopore Sequencing, Algorithms, Chromosomes, Human, Haploidy, HLA Antigens, Humans, Genome, Human, Sequence Analysis, DNA, Benchmarking, Genomics, High-Throughput Nucleotide Sequencing

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Nanopore sequencing and the Shasta toolkit enable efficient de novo assembly of eleven human genomes. https://app.cochat.ai/papers/10.1038/s41587-020-0503-6
The paper
Shafin, K., Pesout, T., Lorig-Roach, R., Haukness, M., Olsen, H. E., Bosworth, C., Armstrong, J., Tigyi, K., Maurer, N., Koren, S., Sedlazeck, F. J., Marschall, T., Mayes, S., Costa, V., Zook, J. M., Liu, K. J., Kilburn, D., Sorensen, M., Munson, K. M., . . . Paten, B. (2020). Nanopore sequencing and the Shasta toolkit enable efficient de novo assembly of eleven human genomes. Nature Biotechnology. https://doi.org/10.1038/s41587-020-0503-6

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.