Chromosomal-level genome assembly of Populus lasiocarpa

Abstract

Populus lasiocarpa, commonly called the Chinese necklace poplar, is a species of poplar native to humid forests of China, and is known for its large leaves that may reach dimensions of 35 × 25 cm. In this study, we generated a high-quality chromosomal-level de-novo assembly and annotation of P. lasiocarpa with a genome size of 419.54 Mb and a gene number of 39,008, which provide an important data support for the conservation and utilization of wild germplasm resources of P. lasiocarpa.

Introduction

Rapid global climate change is posing a main threat to biodiversity. Therefore, revealing the evolutionary history of species, understanding the patterns of genetic diversity, exploring the genetic mechanism of adaptive evolution, and evaluating adaptive capacity of species in the facing with changing environment have laid a theoretical foundation for genetic rescue and informing conservation strategies. Populus lasiocarpa, a species with ring distribution of populations wraping around Sichuan Basin, is unique poplar germplasm resources in China. The geographical barriers and heterogeneous environments constituted by the numerous uplifted mountain surrounding the basin offer ideal materials for studying geographic isolation and genetic mechanisms of adaptive evolution. Notably, a high-quality reference assembly is a key for future related studies. Therefore, we provide a high-quality P. lasiocarpa genome which assembled into chromosome-level here.

Results

Genome assembly and annotation

We first performed K-mer analysis to determine P. lasiocarpa genome size and composition via Illumina sequencing data, revealing an estimated genome size of 451.2 Mb and a heterozygosity rate of 0.6%. To obtain a high-quality genome assembly, we generated 49.95 Gb of Nanopore long-read sequences (∼119×), 36.37 Gb of Illumina short-read data (∼87×), and 61.56 Gb of high-throughput chromosome conformation capture (Hi-C) data (∼149x) for P. lasiocarpa. Using these sequencing data, a 419.54 Mb non-redundant assembly was obtained with the contig N50 size of 9.19 Mb and contig number of 105 after removing redundant sequences and potential contaminated sequences (Table 1). Based on Hi-C read pairs, the assembled contigs were further anchored to 19 pseudo-chromosomes with an average anchoring rate of 99.23% (Table 2). The completeness and accuracy of the assembled genome were validated using benchmarking universal single-copy orthologues (BUSCO) showed that 1,346 complete plant orthologues (97.89%) were recalled (Table 3). The assembly was further evaluated by mapping short reads to the genome, which revealed a mapping rate and single-base accuracy (Depth>=5×) of 97.83% and 99.99%, respectively. Collectively, these results reflected the high level of completeness and reliability of our P. lasiocarpa genome assembly.

Table 1. Statistics of assembled contigs for P. lasiocarpa.

Table 2. Scaffolding of contigs based on Hi-C data.

Table 3. Statistics of BUSCO evaluation for genome assembly.

Genome annotation

We subsequently annotated repetitive elements and protein-coding genes for the final genome assembly. We identified 40.20% of the genome as transposable elements (TEs), which were categorized as long terminal repeat retrotransposons (LTR-RTs) (20.52%), LINE (0.28%) and DNA transposons (15.88%). LTRs formed the most abundant category of TEs, with LTR/Copia and LTR/Gypsy occupying 3.88% and 11.31%, respectively (Table 4).

Table 4. Statistics of repeat sequences in P. lasiocarpa genome.

A total of 39,008 protein-coding genes were annotated with high confidence using a comprehensive strategy that combined homology-based searches, transcriptome-based predictions, and ab initio prediction. The average length for total gene regions, coding sequence (CDS) and intron sequence are 3558.82, 1093.61 and 1930.11 bp, respectively (Table 5). We further evaluated the quality of gene prediction by BUSCO and found that 1,557 out of the 1,614 (96.47%) highly conserved core proteins in the Embryophyta lineage were present in our gene annotation, of which 1300 (80.54%) were single-copy genes and 257 (15.92%) were duplicated. For the remaining conserved genes, 20 (1.24%) had fragmented matches and 37 (2.29%) were missing.

Table 5. Statistics of protein-coding genes

Among the predicted protein-coding genes, 93.96% could be annotated through at least one of the following protein-related databases: Pfam (67.10%), NR (90.62%), Interproscan (87.05%), KEGG (28.76%), Swiss-Prot (69.63%), KOG (79.65%), COG (32.28%), TrEMBLE (91.92%) and GO databases (71.01%) (Table 6).

Table 6. Statistics of protein-coding gene functional annotation

Materials and Methods

Plant materials and genome sequencing

One wild P. lasiocarpa individual (109.74E,30.18N) was selected to harvest fresh young leaves and stems for obtaining high-quality genome assembly. Genomic DNA was extracted from fresh mature leaves using a DNeasy® Tissue Kit (QIAGEN). For the short-read sequencing, 150 bp paired-end libraries with an insert size of 350 bp were constructed and sequenced on Illumina HiSeq X Ten platform. For the long-read sequencing, libraries for Nanopore long reads sequencing were built using large (>20 kb) DNA fragments with the Ligation Sequencing Kit 1D (SQK-LSK109), and sequenced using the PromethION platform (Oxford Nanopore Technologies). For the Hi-C experiment, the libraries were constructed from about 3g of fresh and young leaves and prepared with DpnII restriction enzyme, followed by sequencing on the Illumina NovaSeq platform. To assist gene annotation, total RNAs from fresh young leaves and stems were extracted with CTAB procedure to prepare RNA-seq libraries, which were sequenced on Illumina HiSeq X Ten platform.

Genome assembly and quality assessment

The Illumina short reads were first used to estimate the genome size of P. lasiocarpa via a 17-bp k-mer frequency analysis with Jellyfish (v2.3.0)¹. NextDenovo (v2.0-beta.1, https://github.com/Nextomics/NextDenovo) was then used for the preliminary sequence assembly based on the Nanopore long reads. The raw long reads were first error-corrected via NextCorrect with parameters “reads_cutoff=1k, seed_cutoff=30k”, and then assembled via NextGraph with default parameters. To improve the quality of the assembly, corrected ONT long reads (three rounds) and cleaned Illumina short reads (four rounds) were used to polish assembly by Racon v1.3.1² and Nextpolish v1.0.5³, separately. The redundant sequences were subsequently removed by using perge_haplotigs v1.1.1⁴ and the obtained genome assemblies were checked for DNA contamination by searching against the NCBI non-redundant nucleotide database (Nt) using BLASTN, with an E-value cutoff of 1e-5. Then, BUSCO (v4.0.5, embryophyta_odb10 download at 16-Oct-2020)⁵ with default settings was applied to the assessment of assembly integrity.

The draft assembly was further scaffolded using Hi-C reads. Briefly, the Hi-C reads were filtered by fastp v0.20.0⁶ with same parameters described above. The clean reads were then aligned into the draft assembled sequences using bowtie2 v2.3.2⁷ with parameters ‘-end-to-end, -very-sensitive -L 30’. The mapped Hi-C reads were processed to obtain the valid reads pairs by HiC-Pro v2.11.4⁸. Scaffolds were anchored into 19 pseudo-chromosomes using LACHESIS⁹ with parameters CLUSTER MIN RE SITES=100, CLUSTER MAX LINK DENSITY=2.5, CLUSTER NONINFORMATIVE RATIO=1.4, ORDER MIN N RES IN TRUNK=60, ORDER MIN N RES IN SHREDS=60, and then followed by manual correction.

Gene prediction and functional annotation

Before gene prediction, preliminary TE annotation was performed by EDTA v1.9.3¹⁰ pipeline and TEs annotated as LTR/unknown were re-classified using TEsorter v1.2.5¹¹. The EDTA-constructed TE libraries were applied to mask the whole genome sequences using RepeatMasker v4.10¹².

We integrated three strategies including homology-based prediction, transcriptome-based prediction, and ab initio prediction to predict the protein-coding genes. To perform homology-based prediction, we aligned the protein sequences from six species (Populus trichocarpa, Populus euphratica, Salix brachista, Salix purpurea, Arabidopsis thaliana and Vitis vinifera) to P. lasiocarpa genome by TBLASTN¹³, and parsed the resultant alignments for homolog predictions using Genewise v2.4.1¹⁴. Furthermore, these assembled transcripts based on both genome-free and genome-guided using Trinity v2.8.4¹⁵ were aligned to the genome assembly using PASA v2.4.1 to conduct transcriptome-based predictions. Augustus¹⁶ with default parameters was used to incorporate the homology- and transcriptome-based gene models for ab initio gene prediction. In the end, all above gene models were integrated into a comprehensive gene set using EvidenceModeler v1.1.1¹⁷ which was further updated for three rounds using PASA.

Gene functional annotation was conducted based on BLAST searches with 1e-5 E-value cutoff against four well-known public protein databases: the protein families database (Pfam), the NCBI non-redundant protein database (NR), the interproscan database, the KEGG database, the Swiss-Prot protein database, the Eukaryotic Orthologous Groups of proteins (KOG), the Translated European Molecular Biology Laboratory (TrEMBL) database and the Gene Ontology (GO) database. The putative domains and GO terms of P. lasiocarpa genes were identified using the InterProScan program with default parameters.