Long-read sequencing and de novo assembly of a Chinese genome

Abstract

Short-read sequencing has enabled the de novo assembly of several individual human genomes, but with inherent limitations in characterizing repeat elements. Here we sequence a Chinese individual HX1 by single-molecule real-time (SMRT) long-read sequencing, construct a physical map by NanoChannel arrays and generate a de novo assembly of 2.93 Gb (contig N50: 8.3 Mb, scaffold N50: 22.0 Mb, including 39.3 Mb N-bases), together with 206 Mb of alternative haplotypes. The assembly fully or partially fills 274 (28.4%) N-gaps in the reference genome GRCh38. Comparison to GRCh38 reveals 12.8 Mb of HX1-specific sequences, including 4.1 Mb that are not present in previously reported Asian genomes. Furthermore, long-read sequencing of the transcriptome reveals novel spliced genes that are not annotated in GENCODE and are missed by short-read RNA-Seq. Our results imply that improved characterization of genome functional variation may require the use of a range of genomic technologies on diverse human populations.

Figure 1. Summary of gap filling in GRCh38.

(a) Length distribution of all gaps (stretches of ‘N' in genome sequence) in GRCh38. (b) Length distribution of all gaps that can be fully or partially closed. (c) Violin plots showing the distribution of LINE, SINE, LTR, simple repeat and satellite in closed gaps and in GRCh38. (d) A dotplot showing how a gap on 17p13.3 is closed by a contig in HX1. The plot shows comparison of two sequences and each dot indicates a region of close similarity between them. (e) Genome browser screenshot of the gap region that was closed. The gap is flanked by two contigs that are new in GRCh38 (not carried forward from GRCh37), yet an HX1 associated contig (000850F-001-01) can completely align to flanking regions, therefore filling this assembly gap and revising its length from 718 to 731 bp.

Figure 2. Detection of structural variants by different technologies.

(a) Chromosome ideogram showing large-scale (>1 kb) deletions (blue) and insertions (red) identified from long-read sequencing data. (b) Pie chart showing the distribution of different classes of structural variants identified from long-read sequencing data. (c) Venn diagram showing the overlap of structural variants between HX1, CHM1 and the 1000 Genomes Project for insertions and deletions, respectively. (d) Integrative Genomics Viewer screenshot of the long-read (upper panel) and short-read alignment (lower panel) around an ∼200-kb deletion.(e) Alignment of de novo assembled genome map (blue) to reference genome map (green) where the ∼200-kb deletion occurs. Black vertical lines represent labels for the enzyme recognition site. Contig 2 shows identical label patterns as reference, yet contig 1 contains the deletion. (f) Integrative Genomics Viewer screenshot of long-read (upper panel) and short-read (lower panel) alignment around a 132-bp deletion on KRTAP1-1. This deletion is visually discernible from long-read sequencing, because the coverage is reduced and half the reads contain the deletion in alignments. However, read-depth-based method failed to detect this deletion with short read data. (g) Genome browser screenshot of the region surrounding the 132-bp deletion on KRTAP1-1, demonstrating the presence of simple tandem repeats and the very high GC content of the deletion

Figure 3. Novel gene inferred from Iso-Seq long-read RNA sequencing.

(a) Integrative Genomics Viewer on alignment files generated from Iso-Seq. Over 100 long reads can be mapped to this locus on chr20q13.12 in the GRCh38 assembly. (b) UCSC Genome Browser screenshot on the predicted transcript models. The transcripts are not detected in RNA-Seq data on nine cell lines in ENCODE. This gene is conserved in primates but not in other vertebrate species, and is not in segmental duplication regions or simple repeat regions. (c) PCR validation of the transcript TCONS_0035154 by a primer pair that targeted exons 1 and 5. Several PCR products with different sizes can be detected, representing different isoforms. MC239 is a Caucasian sample and MA296 is an East Asian sample. (d) Sanger sequencing confirmed the splicing events predicted by the Iso-Seq data.

Figure 4. Functional annotation and analysis of the genomic variants in HX1.

(a) Average coverage versus GC contents for 100-bp windows in Illumina data and PacBio data, respectively. The mean and s.d. values are shown. (b) Distribution of PacBio coverage for regions that have ≤5 × coverage in Illumina data. (c) Shared SNVs discovered in HX1, AK1, HuRef, NA12878 and YH. (d) Variant reduction pipeline to identify pathogenic variant; although 20 were annotated as ‘pathogenic' in ClinVar, careful analysis failed to support any one.